Keep Exploring!!!
January 06, 2023
January 05, 2023
Learn how to generate or manipulate images with our DALL·E models
Very Realistic, New variations!!!
Keep Exploring!!!
Questions about my Work ?
Do you code? Are u a Coder / Non-coding manager?
I code whenever I have to do a POC, Evaluating an approach. In a distributed work, when someone is struck if it's Vision / Model Training Code / Clustering / Approach / DB / Design I review/debug. If it's for algo I reuse my past domain inputs to see what are alternatives.
Can you write code and demonstrate capability?
Mind visualizes in a different way, not as methods and packages :)
- For string search, c/c++ for loop comes to mind
- There is regex in python
- semicolon for java statements
- I only remember logic / I StackOverflow it and put syntax
- Do i remember syntax heavily, No
- SQL Server I had spent a considerable time I can write about it anytime
Do you know how Transformer works, can u rewrite it?
- Get back to basics, 1st cut - LSTM you have attention, cell state, memory, forget gate
- Build your blocks - 2nd cut - A bit of encoder (encode input), and a set of decoders added, from a given list pick where attention is maximum
- Code up basics, 3rd cut - Rework / check similar simpler examples/paper on the transformer
- Customize examples - 4th cut - logic/approach/code / keep iterating/building from other repo / customize/working to see if a prototype helps, Still coding / experiment.
Do you know about product management/program management / Data engineering?
Over years I have moved across several roles. You have to collaborate/influence big initiatives/projects. Deep domain knowledge gives us clarity on what solution works and is relevant to our problem statement. I may not have those titles but I have collaborated/delivered with partners in each domain. For bigger initiatives it's Teamwork.
How often do you code?
Out of 10 git codes, I experiment / 1~2 work and for the rest I end up with environmental issues. You keep changing tabs reading products/code/experiments. It's never fixed but it's more about a solution to build and the rest is part of solving the problem.
Everything does not contribute immediately but every type of error adds up to a backlog of 'Oh I have seen it, ' This could be due to this'
Learned knowledge / applied knowledge / inferred knowledge / connected knowledge.
Domain vs Data vs Code vs product balancing all the aspects.
Seniority in software engineering is more than money and promotions.
— Fernando 🇮🇹🇨🇠(@Franc0Fernand0) December 30, 2022
It's about taking on more responsibilities:
• make the right trade-offs
• add value to the business
• give reasonable estimations
• mentor and share knowledgehttps://t.co/ziIiIxWVOr
Why do we Change Roles / Jobs?
- Experiment with different roles/projects
- Impact / Change of Environment
- Learn new things
- Change of place/people
- Growth / Financial benefits
- Money, Work-Life, Win / Failures, Impact, Learning everything matters but at the end of the day pick and choose battles
It's one life, learning never ends, and Nothing comes for free. Keep learning, aware and aligned.
With Covid, I do feel age catching up in my 40s. Pick and choose your battles in Life :). Keep going!!!
How do I learn as a Manager?
- As a manager, if you manage people who can manage themselves then you have time to learn
- As a manager, if you manage people you have to manage then your focus and attention will be divided
- The most painful thing is making someone understand and aware of their weakness
- It's difficult to see other's points of view if we defend, Consulting is all about the right tradeoff time vs effort vs impact.
What are opportunities to Learn / Create Impact?
- Code to prototype solution feasibility
- Learn to review other's domain use cases
- Bridge product vs tech wherever possible
- The lookout where you can resell solutions
- Collaborate for success, Success needs credits shared, and Failure needs credits owned up
We always keep refining our 'best-fit' architecture
Learning is an endless process like layers of an onion, Same approach / never-ending
- Know what you speak
- Code what you speak
- Speak what you code
- Improve on the feedback you hear
- Improve every time when you get an idea
Knowledge is a reflection of your questions, opinions, the code that you deliver, and the design you put on the table. Communication vs Competency vs Consistency all are different skills that you can discover over time.
Keep Exploring!!!
January 04, 2023
AI in manufacturing use cases
Health and safety
- Employee health & safety: incident prevention
- Process safety: advanced alarm analytics
Quality
- Quality inspection in assembly
- Quality assurance/
- defect inspection
- Quality testing
- Quality prediction
Maintenance
- Machine health monitoring:
- predictive maintenance
- Maintenance planning
Production process
- Process optimization
- Line balancing
- Product design and development
- Process parameter optimization
- Production planning/
- decision support
Supply chains
- Future demand and price forecasting
- Supply chain control tower
- Warranty and service management
Energy management
- Energy optimization
- Electricity demand forecasting
- Heating and cooling optimization
Ref - Link
Keep Exploring!!!
Robotics and Vision at Amazon
- Multiple cameras provide a set of eyes on the scene
- 3D scene understanding is generated by multiple camera angles
- Motion algorithms take this understanding of the scene and item identification
- Recognize and estimate the shape of individual items
- Pinch-grasping robot achieved a 10-fold reduction in damage on certain items
- A robot optimized for deployment in a fulfillment center may require suction and pinching
- Depending on the type of object recommend a certain way to handle
Ref - Link
Keep Exploring!!!
Delivery Optimizaton at Amazon
- Local optimizations were tackled first
- Stop consolidation — a chain of addresses linked by a small road segment
- Same-day delivery consolidation
Strategies
- Customer Order and Network Density OptimizeR (CONDOR).
- Simultaneously determining how orders are split into shipments and the sourcing fulfillment center for each shipment.
- Determine the right tradeoff between the levels of complexity and optimality.
- Single warehouse may assemble the order if it has all the items; if not, the order can be split into multiple shipments.
Ref Link
Keep Exploring!!!
The Logistics at Amazon
- Delivery between fulfillment centers (FCs), where inventory is stored, and customers’ doorsteps
- Key factors demand forecasts, cost of acquiring transportation resources, staffing, buffer capacity planning
- Strategy - origin-destination targets, Demand for specific route/load - what percentage of the demand in a given locality should be met by a given fulfillment center?
- Customer demand, region-level demand
- Online controller - scan, label, apply, manifest]
Ref - Link
Keep Exploring!!!
January 02, 2023
January 01, 2023
NLP Tools
Open-source tool for data-centric NLP. Argilla helps domain experts and data teams to build better NLP datasets in less time. =
Haystack is an open source NLP framework that leverages pre-trained Transformer models.
VidGear is a High-Performance Video Processing Python Library that provides an easy-to-use, highly extensible, thoroughly optimised Multi-Threaded + Asyncio API Framework
Lightly is a computer vision framework for self-supervised learning.
An image hashing library written in Python
Deepface is a lightweight face recognition and facial attribute analysis (age, gender, emotion and race) framework for python.
PaddleSeg is an end-to-end high-efficent development toolkit for image segmentation based on PaddlePaddle
Best-of Machine Learning with Python
Keep Exploring!!!
Subscribe to:
Posts (Atom)
