🔥 Quick Notes for ML, DS, MLOps, LLMOps 🔥

It contains interview preparation notes from iNeuron, article links, and others.

Important Concepts:

1. What is the difference between filter, wrapper, and embedded methods for feature selection? Answer
2. 120 Questions. Answer
3. Probability vs. Likelihood. Answer My Fav.: StatQuest
4. Generative and discriminative. Answer
5. ML concepts and code. Answer
6. EM - Expectation-Maximization. Answer
7. Random Forest. Answer
8. Regression - Type of change. Answer
9. Pearson vs Spearman vs Kendall: Stackexchange
10. Gain and Lift Charts. listendata
11. Statistical Hypothesis tests in Python. Jason
12. Machine learning system design. Link
13. A/B Testing. Link, Link
14. Product Questions. Quora
15. Random Forest to Layman. Quora
16. ANOVA, ANCOVA etc. Link
17. ML System Design Template Link
18. PM Technical Concepts Link
19. Trustworthy Online Controlled Experiments Link
20. Product Sense Link
21. Tableau-style User Interface for visual exploration Link

Useful blogs to refer to:

1. Martin Henze (Heads or Tails). Blog
2. Python Snippets. Link
3. PandasVault. Link
4. Python Engineer. Twitter
5. Paired vs Unpaired data: link
6. Data-informed product building: Link
7. Metric: Link, Link,SQL
8. Into to Linear Algebra: Link
9. IMS data sources: Link
10. Predictive model performance check: ListenData
11. Case Study: Link
12. Collection of cases: Link, GAME
13. Gradient Boosting: Link
14. Federated learning: Link, Link2
15. MLOps: Link
16. Mixed Effect Models: Link, Link1
17. ML System feature store: Link
18. Data Science Cheat Sheet: Link
19. Things can go wrong: Link
20. Transformers from scratch Link
21. Dive into Deep Learning Link
22. DL Interview Link
23. DL Rules of Thumb Link
24. ML Forecasting Link
25. MLOps without much Ops Link
26. Rules of Machine Learning by Google Link
27. Product Management for AI Link
28. Feature Engineering and stacking Link
29. Distilled AI Link
30. Leetcode List link
31. There is only one test Link
32. Engineering Practices for DS Link
33. MLE Flashcards Link
34. Time Series Forecasting Link
35. MLStack.Cafe Link
36. Agile data science Link
37. Matt Mochary Method Link
38. Nubank Link

ML System Design:

1. Framework Link
2. Product minded ML design. Link
3. ML Design Link
4. MLE Book Link
5. ML System design Link
6. Full stack deep learning Link
7. Production Machine Learning Problems Link
8. ML System Design Resources Link
9. Metric Question Link
10. Product Matrics Link
11. ML Stack Template Link
12. Patrick Halina - ML Design Link
13. ML Interview Link
14. ML Cheat Sheet Link
15. ML Project Timelines Link
16. ML in Production Link
17. MLOps Paper Link
18. MLOps Questions Link
19. System Design Videos Link
20. Instacart MLOps Link
21. ML Tests Link
22. Operationalizing Machine Learning Link
23. Swirlai Link
24. Twitter Recommendation Link
25. AskItRight Link

Foundation Models:

1. Stanford LLM Link
2. Better Product Search Link
3. Natural Language Processing with Deep Learning Link
4. Chat2Vis Link
5. RAG vs Fine-tuning Link
6. Practical RAG Link
7. LLM Zoomcamp Link
8. What We Learned from a Year of Building with LLMs Link Link
9. LongRAG Link
10. RouteLLM Link
11. ColPali: Efficient Document Retrieval with Vision Language Models Link
12. LLama 3.1 Link
13. 🧠 IncarnaMind Link
14. Mastering LLMs Link
15. LongRAG Link
16. Denser Retriever Link
17. RAG Techniques Link
18. PDF extract Link
19. crawl4ai Link
20. NotebookLM Link
21. Open Data QA Link
22. Contexual RAG Link
23. Long-Context LLMs Meet RAG Link

Useful LinkedIn Posts:

Understand the business context first, don't get over-excited about the tech, and jump into coding too early. When someone asks you for a model, always ask:

👉 why do you need it? 👉 what is your current solution (e.g. what is the baseline to beat)? 👉 who is going to use the predictions and how? 👉 what is the financial impact of the model’s downtime or mistakes? 👉 which metrics do we care about to measure what?

Once you have your answers, back them up with a solid exploratory data analysis, and when you're done, loop in the business team again.

This is a critical moment as your results will translate into 3 potential outcomes:

💡 “Really? This contradicts what I thought. Well, in this case, the ML model doesn’t make much sense anymore”. You are off the hook without a single line of code 🔴 💡 “Ah, interesting. I guess we’ll have to change requirements/scope then.” Course-correct before moving forward 🟠 💡 “This is what I expected. Let’s go ahead”. Greenlight 🟢

Vin Vashishta:

“Next year, we can deliver $X in cost savings and revenue. Last year we delivered B projects resulting in C revenue and D cost savings. We plan to grow that by E%, requiring an F% increase in our total budget to execute.”