About
From small robots to systems that evaluate AI systems.
I am an AI engineer and technical educator based in India. I currently lead AI engineering work at Weights & Biases (part of CoreWeave), focused on coding agents, evaluation harnesses, memory, and improvement loops for agentic systems.
The short version of my work is: build the system, measure it honestly, and teach what survives contact with reality.
Path
Leading a team working on coding agents, internal benchmarks, memory, and applied research at W&B / CoreWeave.
Built production RAG evaluation, LLM and agent integrations, courses, talks, and the evaluation-driven refactor of Wandbot.
Created open-source integrations, technical reports, Kaggle programs, and practical material across deep learning and MLOps.
Studied Electronics and Communication Engineering at Netaji Subhash Engineering College, Kolkata; built autonomous robots and early applied-ML projects.
How I work
I am drawn to the part of AI engineering that is difficult to see in a demo: instrumentation, datasets, failure analysis, evaluation criteria, and the feedback loops that make a system improve. I enjoy moving between implementation and explanation—building a tool, stress-testing it, and then writing or teaching the reusable idea behind it.
That pattern connects most of my work: integrations that expose what a model is doing, evaluations that replace intuition with evidence, and courses that turn months of engineering lessons into a path someone else can follow.
Recognition
- Google Developer Expert in Machine Learning, 2022–2025
- Kaggle Notebooks Master
- Solo silver medal in Kaggle’s 2026 Orbit Wars, placing 132nd among 4,729 teams
- Third place in EngiNX 2018, from more than 19,000 participating teams
The “Enginear” story
When I was ten, a teacher asked what I wanted to become. I said “engineer.” Asked to spell it, I got it wrong, and was told that someone who could not spell the word could not become one.
I kept the misspelling—Enginear—as a reminder that an early mistake is not a verdict. Curiosity, practice, and the willingness to keep building matter much more.