Building LLMs is easy.
Knowing if they’re actually good isn’t.
I wrote The LLM Evaluation Rule Book — a concise guide to evaluating LLMs in the real world, with practical decision-making rules.
Bookmark this👇 https://t.co/uzj885VBXb
LLM evals are more than asking:
“Does this response look good?” 👀
I wrote a practical field guide to building production-grade LLM eval pipelines - from datasets & judges to RAG, agents, regression testing & monitoring.
Bookmark this 👇
https://t.co/izXVLD6ric
Statistics is what turns data into evidence.
It helps you answer:
• Is this result real?
• How uncertain is it?
• Does this relationship actually matter?
A practical field guide to thinking from intuition → inference 👇
https://t.co/kQCq0g4pR5
Data Science isn't:
Python → train model → report metric.
It's:
Ask the right question → collect the right data → quantify uncertainty → test assumptions → interpret carefully → make a decision.
The model is only one piece.
🧵
Most ML math feels disconnected.
But how do they actually fit together?
I tried to build that map:
Data → Linear Algebra → Geometry → Calculus → Optimization →Probability → Learning → Generalization
Serious about ML? Bookmark this👇
https://t.co/wJA1QweMGl
@vijayasankarv@Genpact Here's a link to my resume. I believe I could contribute as an AI Engineer at Genpact. If my profile seems like a good fit, I'd appreciate your consideration.
Cheers, and thanks!
https://t.co/pWgXIQ9GlN
Presenting Practical Approaches to Machine Learning—a 9-part series, each covering a crucial stage of the ML pipeline. Give it a read—I guarantee you'll learn something new!
A Thread ⬇️