AWS Lambda Basics

 Basic Example flow

1. Basic Python Code

2. Create a lambda function

3. Replace function insider handler - change handler - entry point

4. Test the handler

5. Add Trigger to invoker from outside / Gateway

6. Test from Browser

Keep Exploring

Ref - Link

Performance profiling

Happy Learning!!!

Supply Chain Tower - Basics

What is a supply chain tower in retail?

A supply chain tower in retail is a set of integrated processes that enable retailers to manage the flow of goods from the suppliers to the customers. This includes the procurement, storage, distribution, and delivery of products. It also involves the management of inventory, pricing, and customer service. The goal of the supply chain tower is to ensure that the right products are delivered to the right customers at the right time and cost.

What are key components of supply chain tower

• Demand Planning: This involves forecasting and anticipating customer demand, and setting production and inventory levels accordingly.
• Procurement: This involves sourcing, negotiating, and purchasing the necessary materials and components for production.
• Production: This involves the actual manufacturing of the product, and may involve multiple steps and processes.
• Inventory Management: This involves tracking and managing inventory levels, and ensuring that the right products are available when needed.
• Logistics: This involves the transportation and distribution of the finished product to the customer.
• Customer Service: This involves providing customer support and managing customer relationships.

What are different products available for supply chain tower offering?

• Supply Chain Planning: This includes solutions for demand planning, supply planning, inventory optimization, and production planning.
• Supply Chain Execution: This includes solutions for order management, warehouse management, transportation management, and global trade management.
• Supply Chain Analytics: This includes solutions for analytics, business intelligence, and reporting.
• Supply Chain Collaboration: This includes solutions for supplier and customer collaboration, supplier performance management, and product lifecycle management.
• Supply Chain Visibility: This includes solutions for real-time visibility, event management, and tracking.
• Supply Chain Optimization: This includes solutions for supply chain optimization, process automation, and optimization.

What are different commercial products available for supply chain tower offering?

• Oracle Supply Chain Management Cloud: Oracle Supply Chain Management Cloud is a comprehensive suite of cloud-based applications that enable organizations to manage their entire supply chain from end-to-end. It includes modules for demand planning, order management, inventory optimization, and transportation management.
• SAP Ariba Supply Chain Tower: SAP Ariba Supply Chain Tower is a cloud-based platform that helps organizations manage their supply chain processes from end-to-end. It includes modules for demand planning, supply chain analytics, and supplier management.
• JDA Supply Chain Tower: JDA Supply Chain Tower is a comprehensive suite of cloud-based applications that enable organizations to optimize their supply chain operations. It includes modules for demand forecasting, inventory optimization, and transportation management.
• Infor Supply Chain Tower: Infor Supply Chain Tower is a cloud-based platform that helps organizations manage their supply chain processes from end-to-end. It includes modules for demand planning, supply chain analytics, and supplier management.
• Kinaxis RapidResponse: Kinaxis RapidResponse is a cloud-based supply chain management platform that enables organizations to manage their entire supply chain from end-to-end. It includes modules for demand planning, supply chain analytics, and inventory optimization.

What it takes to build supply chain tower open source 

• Develop a clear vision and strategy: The first step in building a supply chain tower open source is to develop a clear vision and strategy for the project. This should include a detailed plan for how the project will be developed, what type of technology will be used, and how the project will be maintained and supported.
• Gather resources: Once the vision and strategy have been established, the next step is to gather the necessary resources to build the supply chain tower open source. This includes gathering the necessary personnel, software, hardware, and other resources needed to develop the project.
• Design the architecture: The third step is to design the architecture of the supply chain tower open source. This includes deciding on the type of technology to be used, the architecture of the software, and the overall structure of the project.
• Develop the software: Once the architecture has been designed, the next step is to develop the software. This includes writing code, testing the software, and ensuring that the software meets the requirements of the project.
• Test the software: Once the software has been developed, the next step is to test the software to ensure that it meets the requirements of the project. This includes running tests, debugging the software, and ensuring that the software is secure and reliable.
• Deploy the software: Once the software has been tested, the next step is to deploy the software. This includes setting up the necessary infrastructure, deploying the software to the production environment, and ensuring that the software is available to users.
• Maintain and support the software: The final step is to maintain and support the software. This includes providing bug fixes and updates, responding to user feedback, and ensuring that the software is secure and reliable.

What is 09 supply chain offering and who are its competitors

09 Supply Chain is an online marketplace for businesses to find and purchase services related to logistics, transportation, and supply chain management. It provides a platform for businesses to compare prices, find the best deals, and connect with logistics providers. Its competitors include Freightos, Flexport, and Shipa Freight.

Top 10 ways to have real time inventory

• Automate inventory tracking with barcode scanners.
• Implement a cloud-based inventory management system.
• Utilize radio-frequency identification (RFID) technology.
• Utilize mobile devices for inventory tracking.
• Utilize inventory management software.
• Utilize automatic data collection (ADC) systems.
• Utilize point-of-sale (POS) systems.
• Utilize cycle counting.
• Utilize automated replenishment systems.
• Utilize automated inventory tracking systems.

Top 10 ways to have real time replienshment planning

• Utilize real-time inventory management systems.
• Analyze customer demand patterns.
• Establish a reorder point.
• Develop a replenishment strategy.
• Automate the replenishment process.
• Monitor supplier performance.
• Utilize forecasting models.
• Utilize safety stock.
• Utilize vendor-managed inventory.
• Monitor inventory levels.

Compare features between Freightos, Flexport, and Shipa Freight, 09 Supply Chain

Freightos:

• Instant online freight quotes and bookings
• Access to global carriers and freight forwarders
• Real-time tracking and visibility
• Automated document generation
• Integrated payments
• Customs clearance
• Insurance
• Freight audit and payment
• Data analytics

Flexport:

• Global freight forwarding
• Customs brokerage
• Warehousing and distribution
• Real-time tracking and visibility
• Data analytics
• Automated document generation
• Integrated payments
• Insurance
• Freight audit and payment

Shipa Freight:

• Global freight forwarding
• Instant online freight quotes and bookings
• Real-time tracking and visibility
• Automated document generation
• Integrated payments
• Customs clearance
• Insurance
• Freight audit and payment
• Data analytics

Keep Exploring!!!

Customer360 - Personal - GPT

Building some fundamentals with GPT :)

What are different third party sources retailers use to discover customers

1. Social media platforms (Facebook, Twitter, Instagram, etc.)

2. Email marketing

3. Affiliate marketing

4. Online advertising (Google Ads, Bing Ads, etc.)

5. Influencer marketing

6. Online review sites (Yelp, TripAdvisor, etc.)

7. Mobile marketing

8. Search engine optimization (SEO)

9. Referral programs

10. Content marketing

What are different customer persona for retail fashion industry

1. The Trendsetter: This customer is always looking for the latest fashion trends and loves to be the first to wear them. They are willing to pay a premium for the latest styles and are always on the lookout for the newest fashion collections.

2. The Bargain Hunter: This customer loves to find the best deals and discounts on fashion items. They are always looking for the best prices and are willing to wait for sales or discounts to purchase items.

3. The Quality Seeker: This customer is willing to pay for quality fashion items that will last them for years. They are willing to invest in higher-end fashion items that are made with quality materials and craftsmanship.

4. The Comfort Seeker: This customer is looking for fashion items that are comfortable and stylish. They are willing to pay for items that are both fashionable and comfortable.

5. The Classic Dresser: This customer is looking for timeless fashion items that will never go out of style. They are looking for items that will last them for years and are willing to pay for quality and timeless fashion items.

What are different customer persona for food consumption dining industry

1. The Foodie: This customer is passionate about food and loves to try new dishes. They are always looking for the latest trends in food and are willing to pay for quality.

2. The Health Conscious: This customer is focused on eating healthy and nutritious meals. They are willing to pay for organic and natural ingredients and are always on the lookout for healthy options.

3. The Budget Conscious: This customer is looking for the best value for their money. They are always looking for discounts and deals and are willing to sacrifice quality for a lower price.

4. The Socialite: This customer loves to go out and socialize with friends. They are looking for a fun atmosphere and are willing to pay for a good time.

5. The Family: This customer is looking for a place to take the family. They are looking for a comfortable atmosphere and are willing to pay for convenience.

What are different customer persona for customer spending

1. The Budget-Conscious Shopper: This customer is focused on getting the best value for their money and is always looking for discounts and deals. They are likely to comparison shop and research products before making a purchase.

2. The Impulse Buyer: This customer is driven by emotion and tends to make purchases on a whim. They are likely to be drawn to flashy displays and attractive packaging.

3. The Loyal Customer: This customer is loyal to a particular brand and will always purchase from that brand regardless of cost or convenience. They are likely to be influenced by loyalty programs and customer service.

4. The Early Adopter: This customer is always looking for the newest products and trends. They are likely to be influenced by word-of-mouth and social media.

5. The Experiential Shopper: This customer is focused on the shopping experience and is likely to be influenced by store layout, customer service, and overall atmosphere.

What are different customer persona for healthcare spending

1. The Young Professional: This customer is a young adult who is just starting out in their career and is looking for ways to save money on healthcare costs. They may be on a tight budget and may be looking for ways to reduce their out-of-pocket expenses.

2. The Family: This customer is a family with multiple members who need to be covered under a healthcare plan. They are looking for a plan that will provide the best coverage for their family at the lowest cost.

3. The Senior: This customer is an older adult who is looking for ways to save money on healthcare costs. They may be on a fixed income and may be looking for ways to reduce their out-of-pocket expenses.

4. The Student: This customer is a student who is looking for ways to save money on healthcare costs. They may be on a tight budget and may be looking for ways to reduce their out-of-pocket expenses.

5. The Self-Employed: This customer is a self-employed individual who is looking for ways to save money on healthcare costs. They may be looking for ways to reduce their out-of-pocket expenses and may be looking for a plan that will provide the best coverage for their needs.

What are different customer persona for bank loans

1. The Young Professional: A young professional who is looking to purchase a home, car, or other large purchase and needs a loan to do so. They are looking for a loan with competitive interest rates and flexible repayment terms.

2. The Small Business Owner: A small business owner who needs a loan to finance their business operations. They are looking for a loan with low interest rates and flexible repayment terms.

3. The Retiree: A retiree who is looking for a loan to supplement their retirement income. They are looking for a loan with low interest rates and flexible repayment terms.

4. The Student: A student who is looking for a loan to finance their education. They are looking for a loan with low interest rates and flexible repayment terms.

5. The Investor: An investor who is looking for a loan to finance their investments. They are looking for a loan with low interest rates and flexible repayment terms.

How to build data across online and offline the data engineering blueprint

1. Identify the data sources: Identify all the available data sources, both online and offline, that you need to build your data engineering blueprint. This includes data from websites, databases, APIs, mobile apps, and other sources.

2. Collect the data: Collect the data from the identified sources and store it in a secure location. Depending on the type of data, you may need to clean and transform the data before storing it.

3. Create a data model: Create a data model that will allow you to store, access, and analyze the data. This model should be designed to accommodate the data sources you have identified.

4. Create a data pipeline: Create a data pipeline that will allow you to move data from the data sources to the data model. This pipeline should be designed to be efficient and secure.

5. Create a data warehouse: Create a data warehouse that will store the data from the data pipeline. This warehouse should be designed to be secure and scalable.

6. Create a data visualization layer: Create a data visualization layer that will allow you to visualize the data in the data warehouse. This layer should be designed to be user-friendly and interactive.

7. Create an analytics layer: Create an analytics layer that will allow you to analyze the data in the data warehouse. This layer should be designed to be powerful and flexible.

8. Create a reporting layer: Create a reporting layer that will allow you to generate reports from the data in the data warehouse. This layer should be designed to be easy to use and customizable.

What are different data engineering challenges to build customer 360

1. Data Collection: Collecting customer data from multiple sources and in different formats.

2. Data Integration: Integrating customer data from multiple sources into a single unified view.

3. Data Quality: Ensuring data accuracy and completeness.

4. Data Analysis: Analyzing customer data to gain insights and uncover trends.

5. Data Visualization: Visualizing customer data to make it easier to understand and interpret.

6. Data Security: Ensuring customer data is secure and protected from unauthorized access.

7. Data Governance: Establishing policies and procedures to ensure data is managed and used responsibly.

Keep Exploring!!!

Fintech - Product Inspirations

This could be part of Navi / Digital Apps

Product Ref - Link

Dashboard to List Customers / Risks (ML Classification)

Risk behavior for repayment (ML / Classification / Forecast)

Digital / Installments / Reminders / Choose your collection Strategy

Keep Exploring!!!

SaaS AI Apps

Primarily the work will be around

• Large-scale Microservices based SAAS enterprise application
• ML code and application development interchangeably
• RESTful APIs for internal and partner consumption
• End-to-end ownership of microservices
• Flask / React-based application

Keep Exploring!!!

Location Analytics

Location Analytics - Learn with GPT

What are top location analytics biggest challenges?

• Data Collection: Collecting accurate and reliable data is one of the biggest challenges in location analytics. This data can be difficult to obtain, especially when it comes to customer data.
• Data Integration: Integrating data from multiple sources can be a challenge. Location analytics requires data from multiple sources to be combined and analyzed in order to gain meaningful insights.
• Data Security: Location analytics requires sensitive customer data to be collected and stored. Ensuring that this data is secure and protected from unauthorized access is a major challenge.
• Data Analysis: Analyzing large amounts of data can be a challenge. Location analytics requires complex algorithms and data mining techniques to be used in order to gain meaningful insights.
• Data Visualization: Visualizing the data in an easy to understand format is a challenge. Location analytics requires data to be presented in a way that is easy to understand and interpret.

What are key location analytics innovative concepts.

• Heat Mapping: Heat mapping is a location analytics tool that uses color-coded maps to visualize the density of data points in a given area. It can be used to identify patterns, trends, and outliers in customer behavior, sales performance, or other metrics.
• Geofencing: Geofencing is a location analytics tool that uses GPS or RFID technology to create virtual boundaries around a geographic area. It can be used to trigger notifications or other automated responses when a customer enters or leaves a designated area.
• Location-Based Targeting: Location-based targeting is a location analytics tool that uses customer data to deliver targeted messages or offers to customers based on their current location. This can be used to drive foot traffic to a store, increase online sales, or promote local events.
• Proximity Analysis: Proximity analysis is a location analytics tool that uses customer data to identify customers who are close to each other. This can be used to identify clusters of customers who may be interested in similar products or services.
• Location-Based Insights: Location-based insights is a location analytics tool that uses customer data to identify trends in customer behavior. This can be used to identify areas of opportunity or areas of risk in a given market.

What are top location analytics trends 

• Predictive Analytics: Predictive analytics is becoming increasingly popular in location analytics, as it allows businesses to anticipate customer needs and behaviors. This helps them create more targeted campaigns and better understand customer preferences.
• Geofencing: Geofencing is a location-based technology that allows businesses to target customers within a specific geographic area. It’s becoming increasingly popular as a way to deliver personalized messages and offers to customers based on their location.
• Location-Based Advertising: Location-based advertising is a form of targeted advertising that uses location data to deliver personalized ads to customers. This helps businesses reach customers with relevant messages and offers, and can be used to increase engagement and sales.
• Real-Time Insights: Real-time insights are becoming increasingly important in location analytics, as they allow businesses to quickly respond to customer needs and trends. This helps them create more effective campaigns and better understand customer behavior.
• Heat Mapping: Heat mapping is a popular location analytics tool that uses data to visualize customer activity and behavior. This helps businesses identify areas of high customer activity and better understand customer preferences.

Location Analytics dataset formats samples key attributes

Location Analytics datasets typically contain a variety of data formats, including geographic coordinates (latitude and longitude), address data, and other location-based information. Common key attributes in these datasets include:

• Location Name: The name of the location, such as a business, park, or landmark.
• Address: The street address of the location.
• Latitude/Longitude: The geographic coordinates of the location.
• Category: The type of location, such as a restaurant, store, or park.
• Visits: The number of visits to the location.
• Time Spent: The amount of time spent at the location.
• Demographics: Information about the people who visit the location, such as age, gender, and income.

What are top startups working on location analytics

• Placed: Placed is a location analytics company that helps marketers understand how their campaigns drive store visits and other real-world behaviors.
• Foursquare: Foursquare is a location intelligence platform that helps businesses understand their customers and drive more foot traffic to their stores.
• PlaceIQ: PlaceIQ is a location intelligence platform that helps marketers understand consumer behavior in the physical world.
• Geofeedia: Geofeedia is a location analytics platform that helps marketers understand how their campaigns are performing in real-time.
• Factual: Factual is a location data platform that helps businesses understand their customers and make better decisions.

Location Analysis

Intro to Indoor Location & Navigation

Indoor Navigation: Complete Data Understanding

Keep Exploring!!!

Tech Trends

Jan to June 2022

Bitcoin, NFT were top trending :)

Explore search interest for ChatGPT, Augmented reality, Metaverse, Non-fungible token, bitcoin by time, location and popularity on Google Trends 

Ref - Link

Next 7 Months - 2023 Trends

ChatGPT has taken lead :) Metaverse is nowhere seen in picture

Ref - Link

Keep Exploring!!!

Multi-Channel Analytics

Multi-Channel Analytics, Patient Pathways, Omni-Channel Segmentation, Territory Design, Customer Targeting, Attribution Modeling, & Predictive Commercial Mix

Multi-Channel Analytics key use cases and features to use

Multi-Channel Analytics can be used to improve customer engagement and understanding across channels, including digital, social, and mobile. Some key use cases for Multi-Channel Analytics include:

1. Understanding customer sentiment across channels - Omni-Channel Segmentation key features

• Segmentation of users into homogeneous groups based on their behavior or interests
• Creation of target audiences for specific products or services
• Optimization of marketing campaigns and content to reach the right audience

2. Identifying customer needs and preferences across channels

Top ways to Identify customer needs and preferences across channels

There are a number of ways to identify customer needs and preferences across channels. Some of the most common methods include customer surveys, focus groups, and interviews.

3. Measuring customer engagement across channels

4. Generating customer insights from multiple channels

5. What are top customer insights from multiple channels?

There are a number of customer insights that can be gleaned from multiple channels, including social media, email, and customer service. Some of the key insights that can be gleaned from social media include customer sentiment, customer feedback, and customer demographics. Email can provide insights into customer engagement, customer service issues, and customer buying behavior. Customer service can also provide insights into customer sentiment, customer service issues, and customer buying behavior.

6. Optimizing customer experiences across channels

There are a few key things that you can do to optimize customer experiences across channels:

• Use customer feedback to improve your product - Use customer feedback to improve your product. This can be done through surveys, focus groups, or customer feedback forums.
• Use social media to connect with customers - Use social media to connect with customers. This can be done through Facebook, Twitter, and other social media platforms.
• Use customer service to resolve issues - Use customer service to resolve issues. This can be done through phone, email, or live chat.
• Use marketing to attract new customers - Use marketing to attract new customers. This can be done through advertising, social media, and other marketing channels.

Keep Exploring!!!

Another fancy / smart way to apply NLP

Text to Shop’ feature, Conversational commerce, or shopping via text, Text Walmart the items they want to purchase from either their local stores or Walmart.com. The blog lists issues/bugs with responses. In a way more engagement.

Improvements - Converge options to refine

• Re-order from previous orders
• Reuse quantity from previous orders
• Select by brand 
• Select by category

I would with a combination of Text, Yes / No Questions, and Nail down to category / Section. List by size/category. 

 Walmart 'fancy' NLP way to shop link

Keep Exploring!!!

Hyper-Local - Emerging Retail Trend

• Hyper-Local Food is the conscious decision to buy food that is grown very close to where you are currently living. 
• These foods will typically be grown and created by your local farmer or rancher.
• Incentivize your local farmers and small businesses
• Definition of real food is no ingredients labels - just the single source ingredient itself.

Ref - Link

Keep Exploring!!!

GPT to learn some more ML optimisation / causal inference

What is numerical optimisation, How it is different from machine learning. List some usecases problems in numerical optimisation ?

Numerical optimization is the process of finding the values of input parameters that minimize or maximize a given objective function. It is different from machine learning in that it does not involve training a model with data, but instead relies on mathematical algorithms to find the best solution.

Some use cases of numerical optimization include:

• Portfolio optimization
• Supply chain optimization
• Vehicle routing
• Facility location
• Network design
• Scheduling
• Parameter estimation
• Image registration
• Model calibration
• Optimization of control systems

What are top 10  python packages for numerical optimization

• NumPy 
• SciPy 
• Scikit-Optimize 
• PyOpt 
• PyGMO 
• CVXOPT 
• Pulp 
• Pyomo 
• DEAP 
• Heuristic Optimization Algorithms (HOALib)

List some constraint for windmill turbine based on temperature, sunlight, humidity, windspeed with synthetic data python code pulp optimization solution

constraints = [

    # Temperature

    pulp.lpSum([temp_var[i] for i in range(len(temp_var))]) <= max_temp,

    pulp.lpSum([temp_var[i] for i in range(len(temp_var))]) >= min_temp,

    # Sunlight

    pulp.lpSum([sunlight_var[i] for i in range(len(sunlight_var))]) <= max_sunlight,

    pulp.lpSum([sunlight_var[i] for i in range(len(sunlight_var))]) >= min_sunlight,

    # Humidity

    pulp.lpSum([humidity_var[i] for i in range(len(humidity_var))]) <= max_humidity,

    pulp.lpSum([humidity_var[i] for i in range(len(humidity_var))]) >= min_humidity,

    # Windspeed

    pulp.lpSum([windspeed_var[i] for i in range(len(windspeed_var))]) <= max_windspeed,

    pulp.lpSum([windspeed_var[i] for i in range(len(windspeed_var))]) >= min_windspeed

]

List some usecases problems in operations research ?

• Vehicle Routing Problem
• Scheduling Problem
• Network Flow Problem
• Knapsack Problem
• Integer Programming Problem
• Queuing Theory Problem
•  Linear Programming Problem
•  Resource Allocation Problem
•  Facility Location Problem
•  Project Scheduling Problem

List some usecases problems in causal inference ?

• Estimating the effect of a treatment on an outcome of interest
• Estimating the effect of a policy change on a population
• Estimating the effect of a marketing campaign on sales
• Estimating the effect of a new product on customer retention
• Estimating the effect of a change in pricing on customer demand
• Estimating the effect of a change in customer service on customer satisfaction
• Estimating the effect of a change in advertising on brand awareness
• Estimating the effect of a change in the labor market on wages
• Estimating the effect of a change in government regulations on economic growth
• Estimating the effect of a change in the environment on public health

What is casual inference in layman terms

Casual inference is a type of reasoning that allows us to draw conclusions based on our observations. In layman's terms, it is the process of making assumptions or guesses about something based on what we know or have seen. For example, if you see someone walking down the street wearing a suit and carrying a briefcase, you can infer that they are likely on their way to work.

List some usecases problems in Bayesian modelling ?

• Estimating the probability of a medical diagnosis given a set of symptoms.
• Predicting the probability of a customer purchasing a product given their browsing history.
• Estimating the probability of a machine failure given a set of sensor readings.
• Predicting the probability of a customer defaulting on a loan given their financial history.
• Estimating the probability of a terrorist attack given a set of intelligence data.
• Predicting the probability of a stock price movement given market data.
• Estimating the probability of a natural disaster given weather patterns.
• Predicting the probability of a customer churning given their usage data.

What is difference between causal interference vs AB testing

Causal inference is a method of determining the cause-and-effect relationship between two variables. It is used to identify the cause of a particular outcome. It is based on the assumption that there is a causal relationship between the two variables.

AB testing is a method of comparing two versions of a product or service to determine which one performs better. It is used to determine which version of a product or service is more effective. It is based on the assumption that the two versions are statistically different.

List some usecases problems in differentiable programming ?

• Optimizing hyperparameters in neural networks
• Automatically generating neural network architectures
• Automatically generating code from a given specification
• Automatically tuning the parameters of a given model
• Automatically generating data pre-processing pipelines
• Automatically generating feature engineering pipelines
• Automatically generating loss functions
• Automatically generating optimization algorithms
• Automatically generating regularization techniques
• Automatically generating distributed training strategies

Bayesian modeling is generative probability based on historical data, Is it correct ?

Yes, that is correct. Bayesian modeling is a type of generative probability model that uses historical data to make predictions about future events.

When can we use differentiable programming , Does it have advantage over deep learning

Differentiable programming can be used when a problem requires the use of a combination of symbolic and numerical computation. It has the advantage of being able to combine the flexibility of symbolic computation with the power of deep learning. This allows for more complex models to be created and trained, which can lead to better results. Additionally, differentiable programming can be used to create models that are more interpretable and easier to debug.

Keep Exploring!!!

AI vs Natural Intelligence

AI - any device that does things that associate with human intelligence
Natural Intelligence - Imagination, creativity, fantasy, intuition, problem-solving. AI and natural intelligence are two different things, each supplies a lack of other 
Human Interpretation is leaps ahead of Machine Interpretation

Right Use Case

With the Right context / matured model, Tesla Safety Stats - Q3 2022 - In the 3rd quarter, we recorded one crash for every 6.26 million miles driven in which drivers used Autopilot technology.

Purposeful Bound to Fail 

In some cases, where machines can fail, for example in a classification problem

Selecting the Right Use Cases

ML is more of a collaborative effort. Business needs - You spot it, and you can demonstrate competency to get a buy-in to implement. Alignment is critical to understand the short-term/long-term impact. You may not be immediately right but eventually, you may be right

Spotting what others miss needs all three views - domain, data, and ML opportunities.

Datasets Search

• https://datasetsearch.research.google.com/
• https://github.com/awesomedata/awesome-public-datasets
• https://archive.ics.uci.edu/ml/index.php
• https://www.kaggle.com/datasets
• https://msropendata.com/

• Customer related
• Operations related
• Risk Avoidance

Ref - Link

Ref - Link

Keep Exploring!!!

Docker Custom Examples

Some minor fixes working with git project

Keep Exploring!!!

Best practice for container-based deployment system 

• Use a container-orchestration system such as Kubernetes or Docker Swarm to manage and deploy your containers.
• Use a container registry such as Docker Hub or Quay to store and manage your container images.
• Use a continuous integration system such as Jenkins or Travis CI to automate the build and deployment of your containers.
• Monitor your containers and applications using tools such as Prometheus or Grafana to ensure they are running optimally.
• Use a service mesh such as Istio or Linkerd to manage the communication between your services.
• Use a logging and monitoring system such as ELK or Splunk to track the performance of your containers and applications.
• Use a security scanning tool such as Twistlock or Aqua Security to ensure your containers are secure.
• Use a configuration management system such as Ansible or Chef to manage the configuration of your containers.
• Use a deployment automation tool such as Helm or Terraform to automate the deployment of your containers.
• Use a cloud provider such as Amazon Web Services or Google Cloud Platform to host your containers.

Keep Exploring!!!

ChatGPT is 'not particularly innovative'

Good Read - Post1

• ChatGPT is 'not particularly innovative,' and 'nothing revolutionary', says Meta's chief AI scientist.
• The ChatGPT program is a case less of scientific breakthroughs than it is an instance of decent engineering, said LeCun

My Views

• Yes it is a gradual evolution from CNN - RNN - Transformer - Large Language Model
• Earlier we had next character prediction, Now we have sentence level 2048 token
• Alexa is a simple Q&A Bot vs Boston Dynamic Robots. Solutions need to solve the problem no need to be revolutionary.
• Facebook had data but never came up with this LLM :)
• GPT is lead to hybrid variations of chatbots, Alexa, Edge Devices, and Personalized Assistants.

Using GPT for Facts / Data related note retrieval is much better than asking it to solve a maths problem. Remember GIGO (Garbage In, Garbage Out), It may not be able to reason or solve but it can give you templates, formulas for you to get started, We do not read to remember the syntax but focus on solutions.

ChatGPT passes MBA exam given by a Wharton professor

• Amazing job at basic operations management and process analysis questions including those that are based on case studies
• Remarkably good at modifying its answers in response to human hints
• Remarkable ability to automate some of the skills of highly compensated knowledge workers in general and specifically the knowledge workers in the jobs held by MBA graduates including analysts, managers, and consultants

Why ChatGPT was able to create impact ?

• Solving a Niche problem
• Unparalleled user experience
• Plug and Play infra - API
• Human in Loop 
• Reinforcement learning along with Transformers

"Original Thinking / Looking at different way hybrid ideas not just one way" matters

"Perfect vs Good Enough Solution"

Ref  - Link

Good one, 

We can become better writers with ChatGPT

• Knowledge pointers
• Blending creative thought 
• Add your own perspective for the insight
• #Google democratized information. With #ChatGPT #Reasoning / #knowledge / #Facts is getting democratized. 

Why didn't DeepMind build GPT3?

• Organisational-level risk taking involved in building GPT3 - Brand Impact
• Build systems that appear to have intelligence when interacted with.

Keep Exploring!!!

2023 Ideas

Key observations from post

• AI Adoption: More *Fun* Than You Think - AI are about enabling transformation between modalities, whether speech-to-text, text-to-images, or images-to-videos.
• Next Gen tools - hybrid world - beyond Twitter or TikTok, or in a Discord group
• New Channels of Supply Chain - Products sourced directly from manufacturers, Products sourced directly from multiple manufacturers in shared fashion, Products routed from retailers with overstock/return
• Modern Bio Software Stack - Tool makers will focus on making the best tools, 300 biotechs trade at market capitalizations below the cash on their balance sheet
• GPT to overtake / replace human basic discussions - ChatGPT to take input and deliver a near-human-quality credit counseling experience changes all of this, “text to SQL queries” or, eventually, “text to excel modeling”
• Banking - business banking into the digital age - KYC/AML, data APIs, bank accounts as a service, issuer/processors, Fintechs Are Becoming Banks

Big 2023 Ideas

My picks

• More LLM in 2023
• Reduced Training Costs
• GAN creating new opportunities
• EV low-cost battery to fuel more adoption

Keep Thinking!!!

ChatGPT - Food Reviews - Pointers - Examples

List NLP use cases for a food review website

• Sentiment Analysis: Analyzing customer reviews to determine the overall sentiment of the customer towards the food item.
• Text Classification: Categorizing customer reviews into different categories such as positive, negative, neutral, etc.
• Named Entity Recognition: Identifying entities such as ingredients, dishes, restaurants, etc. from customer reviews.
• Topic Modeling: Identifying topics from customer reviews to understand what customers are talking about.
• Summarization: Summarizing customer reviews to provide a concise overview of the customer’s opinion.
• Text Clustering: Grouping customer reviews into different clusters based on their content.
• Automated Response Generation: Generating automated responses to customer reviews based on the sentiment of the review.

List NLP use cases to propose content for users of cooking website

• Recipe Recommendations Based on User Preferences
• Automated Meal Planning
• Automated Grocery Shopping List Generation
• Automated Ingredient Substitution
• Automated Recipe Search
• Automated Recipe Rating
• Automated Recipe Summarization
• Automated Cooking Instructions Generation
• Automated Food Image Recognition
• Automated Food Nutrition Analysis

List different features to collect for Recipe Recommendations of food cooking and review website

• Type of cuisine 
• Ingredients 
• Cooking time 
• Number of servings 
• Dietary restrictions 
• Level of difficulty 
• User ratings 
• Number of reviews 
• User-generated tags 
• Nutritional information 
• Allergen information 
• Price range 
• Cooking methods 
• Region of origin 
• Popularity of recipe 
• User preferences 
• Popularity of ingredients

List different features for Automated Recipe Rating from recipe cooking website

• Taste Rating: Users can rate the taste of the recipe on a scale from 1-5.
• Difficulty Rating: Users can rate the difficulty of the recipe on a scale from 1-5.
• Preparation Time: Automated system can calculate the time required to prepare the recipe.
• Nutrition Rating: Automated system can calculate the nutrition value of the recipe.
• User Reviews: Users can give their reviews and ratings on the recipe.
• Popularity Rating: Automated system can calculate the popularity of the recipe based on the number of views and downloads.
• Ingredient Rating: Users can rate the ingredients used in the recipe on a scale from 1-5.
• Cost Rating: Automated system can calculate the cost of the ingredients used in the recipe.

List key features for Recipe Recommendations of food cooking review website based on seasality

• Automated recipe recommendations based on user preferences and dietary restrictions.
• Ability to search for recipes by ingredients, cuisine, and dietary restrictions.
• Ability to filter recipes based on user ratings and reviews.
• Ability to save favorite recipes and create custom recipe collections.
• Ability to share recipes with friends and family.
• Ability to access recipes from multiple sources, including user-generated content.
• Ability to access recipes in multiple languages.
• Ability to access seasonal recipes and special holiday recipes.
• Ability to access recipes with step-by-step instructions and photos.
• Ability to access nutritional information for each recipe.

List features to engage active users, occasional users of food review website

• Gamification: Offer rewards such as badges, points, and leaderboards to encourage users to post more reviews.
• User Profiles: Allow users to create profiles and follow other users to see their reviews and ratings.
• Social Sharing: Allow users to share their reviews and ratings on social media platforms.
• Recommendation Engine: Use algorithms to recommend restaurants and dishes based on user preferences.
• Personalized Feeds: Create personalized feeds for each user based on their interests and preferences.
• Incentives: Offer incentives such as discounts or freebies for users who post reviews.
• Contests: Hold contests and giveaways to encourage users to post more reviews.
• User Reviews: Allow users to rate and review other users’ reviews.
• Commenting: Allow users to comment on reviews and ratings.
• User Groups: Create user groups to discuss restaurants and dishes.

How do we plug seasonality and recent activity in food recipe recommendation website with feature list and algorithms

Feature List:

• User Profile: Collect user preferences and dietary restrictions to tailor recipe recommendations.
• Seasonal Ingredients: Use seasonal ingredients to suggest recipes that are more relevant to the current season.
• Recent Activity: Track user activity and use it to suggest recipes that are similar to what the user has recently viewed or cooked.
• Popularity: Suggest recipes that are popular among other users.
• Location: Use user location to suggest popular recipes in the region.

• Algorithms:
• Collaborative Filtering: Use collaborative filtering to recommend recipes based on user similarities.
• Content-Based Filtering: Use content-based filtering to recommend recipes based on user preferences and dietary restrictions.
• Hybrid Algorithms: Combine collaborative filtering and content-based filtering to create a more personalized recommendation system.
• Natural Language Processing: Use natural language processing to identify user intent and suggest recipes accordingly.

How do we plug seasonality, age, food interest, recent purchase and recent activity in food recipe recommendation website with feature list 

• Seasonality: Utilize seasonality to recommend recipes that are popular during the current season. For example, in the summer, recommend recipes that include seasonal fruits and vegetables.
• Age: Utilize age to recommend recipes that are appropriate for the user's age group. For example, if the user is a teenager, recommend recipes that are easy to make and require minimal ingredients.
• Food Interest: Utilize food interest to recommend recipes that the user is interested in. For example, if the user is interested in Italian cuisine, recommend recipes that are Italian-inspired.
• Recent Purchase: Utilize recent purchase to recommend recipes that use ingredients the user has recently purchased. For example, if the user has recently purchased a certain type of cheese, recommend recipes that use that cheese.
• Recent Activity: Utilize recent activity to recommend recipes that the user has recently viewed or interacted with. For example, if the user has recently viewed a certain type of recipe, recommend similar recipes.

What coupons, discounts features we can promote on food review site to promote more user engagement

• Loyalty Programs: Offer loyalty programs that reward customers for their repeat business. This could include discounts on future orders, free items, or other incentives.
• Referral Programs: Offer referral programs that reward customers for referring their friends and family to your food review site.
• Coupons: Offer coupons for discounts on orders or free items.
• Discounts: Offer discounts for orders over a certain amount or for certain types of customers (e.g. students, seniors, etc.).
• Special Deals: Offer special deals or promotions on certain days or times of the week.
• Contests: Hold contests or giveaways that reward customers for engaging with your food review site.
• Rewards Programs: Offer rewards programs that reward customers for their loyalty and engagement.
• Social Media Promotions: Promote your food review site through social media channels such as Facebook, Twitter, and Instagram.
• User Reviews: Encourage customers to leave reviews on your food review site. This will help to build trust and credibility with potential customers.
• Newsletter Subscriptions: Offer newsletter subscriptions that provide customers with updates on new products, discounts, and other promotions.

Ideas

• Find insights based on past seasons - Recipe, Ingredients
• Find insights based on holiday seasons - Recipe, Ingredients
• Promote a mix of patterns from past seasons
• The hybrid mix of Events in the timeline (Start Season, Peak Season, End Season)
• User Related groups based on ingredients, age, location, views, follows, likes, shares

Link

• Using a data-driven approach, they segment demand rather than consumers to identify not only what consumers want, but also where, when, why, and how they want it
• Tapping into social media, Chobani obtained real-time information on what kinds of yogurt their consumers want and when they want to eat it. 
• Facebook ads, as a cost-effective way to target consumers demographically, geographically, and psychologically. 
• Top companies charge a premium price that is commensurate with the value of the innovations they make to their product

Incumbents to disruptors: Adopting the start-up culture for innovation

• Free samples Natural
• Sustainable Healthier
• More Effective
• Value-for-money
• Global name
• Discounts

What are Metrics to measure for relevance of personalization

1. Conversion Rate: Measuring the rate at which personalized content leads to conversions can help determine the effectiveness of personalization.

2. Engagement Rate: Measuring the rate at which users interact with personalized content can help determine the relevance of the content.

3. Click-Through Rate: Measuring the rate at which users click on personalized content can help determine the effectiveness of personalization.

4. Time on Site: Measuring the amount of time users spend on a site after being exposed to personalized content can help determine the relevance of the content.

5. Bounce Rate: Measuring the rate at which users leave a site after being exposed to personalized content can help determine the relevance of the content.

Ref - Link

How to we improve engagement on personalized content

1. Make sure the content is relevant to the user.

2. Use data to create personalized content.

3. Use visuals to create a more engaging experience.

4. Ask questions and invite users to interact with the content.

5. Offer incentives for engaging with the content.

6. Utilize social media to reach a larger audience.

7. Use A/B testing to determine what content works best.

8. Monitor user feedback and adjust content accordingly.

9. Encourage user-generated content.

10. Track user engagement and use the data to improve content.

How to increase engagement without personalization in food review site

1. Create a rewards program: Offer customers rewards for leaving reviews, such as discounts or free products. This will encourage customers to leave more reviews and increase engagement.

2. Offer incentives for sharing reviews: Offer customers incentives for sharing their reviews on social media or other platforms. This will help spread the word about your food review site and increase engagement.

3. Host contests: Host contests or giveaways on your food review site. This will encourage customers to leave reviews and increase engagement.

4. Feature customer reviews: Feature customer reviews prominently on your food review site. This will help customers feel like their opinions are valued and increase engagement.

5. Ask customers for feedback: Ask customers for feedback on their experience

How to propose content user has never seen and target engagement

1. Create Unique Content: Create content that is unique and stands out from the rest. This could be in the form of an infographic, a video, an interactive quiz, or a blog post.

2. Utilize Social Media: Leverage social media platforms to reach a wider audience and increase engagement. Share your content on multiple platforms and engage with users who comment or like your posts.

3. Use Targeted Ads: Utilize targeted ads to reach users who are likely to be interested in your content. This could be done through Google Ads, Facebook Ads, or other platforms.

4. Offer Incentives: Offer incentives to users who engage with your content. This could be in the form

What are in organic ways to increase personalization engagement

1. Use personalized emails: Use personalized emails to send targeted messages to customers based on their interests, purchase history, and other data.

2. Personalize content: Create content that is tailored to the individual user’s interests, needs, and preferences.

3. Use segmentation: Segment customers into different groups based on their interests, purchase history, and other data.

4. Leverage user-generated content: Encourage customers to share their experiences with your product or service.

5. Offer personalized discounts: Offer discounts or special offers based on customer data.

6. Utilize customer feedback: Use customer feedback to improve your product or service and create a more personalized experience.

What algorithms quora uses to Amplify high engaging content

1. Natural Language Processing (NLP): Quora uses NLP to analyze user-generated content and identify topics of interest. This helps them to identify content that is likely to be engaging to their users.
2. Machine Learning: Quora uses machine learning algorithms to identify patterns in user-generated content. This helps them to identify content that is likely to be engaging to their users.
3. Recommendation Algorithms: Quora uses recommendation algorithms to suggest content to users based on their interests and past behavior. This helps them to amplify content that is likely to be engaging to their users.
4. Social Network Analysis: Quora uses social network analysis to identify influential users and content that is likely to be engaging to their users.
5. Sentiment Analysis: Quora uses sentiment analysis to identify content that is likely to be engaging to their users based on the sentiment of the content.

What algorithms quora uses to reduce low engaging content

• Content Quality Score: Quora uses a content quality score to identify low-engaging content. This score is based on factors such as the number of views, upvotes, and comments a post has received. Posts with a low content quality score are more likely to be filtered out or marked as low-engaging content.
• Natural Language Processing: Quora uses natural language processing to identify posts that are not relevant to the topic or contain offensive language. Posts that are flagged as low-engaging content are removed from the platform.
• Machine Learning: Quora uses machine learning algorithms to identify low-engaging content. The algorithms analyze user behavior and engagement levels to determine which posts are not engaging enough to be featured on the platform.
• User Feedback: Quora also uses user feedback to identify low-engaging content. Users can flag posts as low-engaging, and Quora will use this feedback to determine which posts should be removed from the platform.

What algorithms quora uses to clean up contents not engaging

• Natural Language Processing (NLP): Quora uses NLP to detect and remove spam, offensive language, and other inappropriate content.
• Machine Learning: Quora uses machine learning algorithms to identify content that is not engaging and remove it from the platform.
• Text Analysis: Quora uses text analysis to identify and remove content that is not relevant to the topic or discussion.
• Content Moderation: Quora uses a team of moderators to review content and remove any content that is not appropriate for the platform.
• Automated Filtering: Quora uses automated filters to detect and remove content that violates its terms of service.

Key Analysis

• Top Writes
• High-Quality Writers
• Low-Quality Writers
• High content features
• Engagement across ages
• Cusines liked across ages
• Regions with engagement
• Time to visit
• Time spent on weekday/weekend
• Time spent over holidays / before holidays
• Response for new launches
• Rank articles by views, upvotes, and comments
• Identify posts that are not relevant
• Determine which posts are not engaging
• Receipt - Novel, Known, Experiment
• Categorize as new recipe, existing but improvements, cook in 5 mins
• Remove poor content in every category
• Re-rank articles to suit to date / season / weekday / weekend
• Custom word embedding created for different food persona
• A stop list or custom choices to avoid any strong dislikes

Why Are Online Recipes Always So Damn Long-Winded?

• Some people love the stories
• SEO (Search Engine Optimization)
• It’s a business
• Recipes aren’t copyrighted

How to ARCHITECT a search engine like Google Search

• Feature creation
• Entity Extraction
• Keyword Matching
• Custom word to vec creation of receipe, food, entity
• A/B Experiments

Possible metrics include

• 7-day new post rates
• 7-day new Upvotes / Comments
• 30-day Sample Conversion rate
• 90-day retention rate/user churn
• 90-day annual revenue per user

Expected Revenue (XR), XR is meant to measure this.

• Variety of factors
• User activity such as Posting recipe, Upvotes, Replies  
• Utilizing the samples
• Inviting friends to join the group
• User location activity Engagement

Yandex Lessons

• Yandex has a RankBrain analogue called MatrixNet
• Yandex also uses PageRank (almost the same as in Google);
• A lot of text algorithms are the same
• 1,922 ranking factors

Key Lessons

• Link age is a ranking factor
• Traffic and % of organic traffic affect rankings
• Things like CTR, last click, time on site, and bounce rate impact rankings
• They take the average position for all your keywords into account. Does that imply that focusing on fewer, high-ranking KWs is better than going for more KWs (with lower positions)?
• Newer pages and recently updated pages have better ranking (no big surprise there)
• More people search for your brand, the higher your (other) pages rank.
• If people bookmark your page, it sends a good signal (probably trust or authority of thoroughness, etc.) that impacts ranking
• Direct traffic is good, If all your site is getting is organic traffic, it may
• look suspicious.
• If you publish low quality content, it can negatively impact your entire site ranking.
• Wikipedia has a special ranking factor
• Traffic from Wikipedia is a ranking factor.
• Backlinks from main pages are more important than from internal pages.
• Special ranking factors for short videos (TikTok, Shorts, Reels)
• Embedded videos are good for rankings, but broken embed videos are bad.
• JS from Google Analytics is a ranking factor (if you use GA, it's good)
• Host reliability is a ranking factor (40x/50x errors)

Reddit Recommendation Inspirations

• Candidate Post Generation - Submissions from the past 24 hours, and filter it through criteria intended to tell us what each user might enjoy
• Community subscriptions: each community you’ve joined
• Similar communities: communities similar to those you have joined (currently we use semantic similarity)
• Onboarding categories: categories you said they were interested in during onboarding (like “Animals & Awws” or “Travel & Nature”)
• Recent communities: communities that the user visited in recent days
• Popular and geo-popular: Posts that are popular among all redditors, or among redditors in their local area (only if permitted in app settings)

Filtering Criteria for Posts

• Every post we show on Reddit must meet a quality and safety threshold, so on the Best Sort we remove posts from the list that we think might be:
• Spam, deleted, removed, hidden, or promoted
• Posts the user has already seen
• Posts from subreddits or topics that the user asked we show less of
• Posts the user has hidden
• Posts from authors the user has blocked

ML Model - “features” for each candidate post.

• Post votes: The number of votes on the post. The magic of Reddit is that it is primarily curated by redditors via voting. This remains at the core of how Reddit works.
• Post source: How we found this post (subscriptions, onboarding categories, etc.)
• Post type: The type of the post (text, image, video, link, etc.)
• Post text: The text of the post
• Subreddit: Which subreddit the post is from, and the ratings, topics, and activity in that subreddit (for more on Ratings and Topics read this).
• Post age: The age of the post (we value giving you a “fresh” Home feed)
• Comments: Comments and comment voting
• Post URL: The URL the post links to, if the post is a link post
• Post flairs: Flairs and spoiler tags on the post

We could rank posts assigned based on views and use it for new posts ranking.

Machine Learning-Powered Search Ranking of Airbnb Experiences

Building signals based on which we will rank: In Stage 1 of our ML model, we decided to rank solely based on Experience Features. In total we built 25 features, some of which were:

Customizing features for Food Review

• Experience duration (e.g. 1h, 2h, 3h, etc.)
• Product Price and Price-per-hour
• Category (e.g. cooking class, music, surfing, etc.)
• Reviews (rating, number of reviews)
• Number of similar items samples used (last 7 days, last 30 days)
• Conversion of coupons (e.g. 60%)
• Click-through rate

Ref - Link

The most commonly observed gamification elements were:

• Levels or progress feedback
• Points or scoring
• Rewards or prizes
• Narrative or theme
• Personalization
• Customization

The least commonly observed gamification elements were:

• Artificial assistance
• Unlockable content
• Social cooperation
• Exploratory or open-world approach
• Artificial challenge
• Randomness

Moving from Tactical Testing to Strategic Experimentation and Personalization

Understanding the difference between personalization and simple A/B testing

The mechanics of games are often reproduced unimaginatively, making the whole experience not fun.

It's important that people actually care about the rewards they receive! 

You should always be upfront about what you are designing with your users.

Gamification isn’t meant to be about tricking you, it’s meant to help you get you on your way, in an ethical way.

Keep Exploring!!!

Web3 slowdown

Some key cut off was Virtual, Mixed Reality, and HoloLens. The hype around the metaverse is coming to reality. GPT in fact is much more realistic than metaverse. Now we see a traction of shift. " it paints a picture of a company that perhaps no longer believes in virtual reality."

Microsoft has laid off entire teams behind Virtual, Mixed Reality, and HoloLens

Google’s Fuchsia and Area 120 see significant cuts in layoffs

Key points

• Nest Hub series and is poised to be used in an upcoming device
• Reply as a smart response toolbar appended to notifications on Android (GPT has killed this I guess)
• Collection of touchscreen games from Area 120’s GameSnacks to play on Android Auto.

I have not seen a great move in Edge Analytics / Google Coral TPU. There was not a great breakthrough in terms of AI adoption / Breakthru models. 

Working on Great Tech vs Delivering Value vs Solving use cases

Keep Exploring!!!

GAN - Observations

Reading about GAN is easier but sad not to get any output after 2000 Epochs.

CycleGAN - CycleGAN is a model that aims to solve the image-to-image translation problem

After 2000 Epochs :( :(. Knowing is 10%, Experimenting is 50%, and Mastering is 40%. Always experiment.

MnistGAN, Loss in CycleGAN

How the loss is calculated while training?

Adversarial Loss: We apply Adversarial Loss to both the Generators, where the Generator tries to generate the images of it's domain, while its corresponding discriminator distinguishes between the translated samples and real samples.

Cycle Consistency Loss: It captures the intuition that if we translate the image from one domain to the other and back again we should arrive at where we started. Hence, it calculates the L1 loss between the original image and the final generated image,

CycleGAn Experiments - Implementing CycleGAN

Image to Image Translation using CycleGANs with Keras implementation

Experimented this code - Code Example

Keras-GAN, Conditional GAN

6 GAN Architectures

• Transforming an image from one domain to another (CycleGAN),
• Generating an image from a textual description (text-to-image),
• Generating very high-resolution images (ProgressiveGAN) and many more

Loss Notes, pixelwise MSE loss

• CycleGAN Tutorial from Scratch: Monet-to-Photo
• CycleGAN and pix2pix in PyTorch
• CycleGAN-PyTorch

Keep Exploring!!!