Machine learning · systems · product
Practical machine learning tools and products, shaped by evaluation, reliability, and real-world constraints.
TL001 train.py:18
Scaler fitted before the train / test split
↳ Fit inside a pipeline after splitting
13
ML checks
0
runtime deps
local
source only
A small example of the problems I like to solve.
Selected work
A few problems I’ve worked on, from ML reliability to document and speech systems.
Open-source developer tool
01An offline Python checker for ML bugs that run without errors: data leakage, training-mode mistakes, incorrect loss inputs, and more. Includes agent skills that help prevent the same issues.
13 checks · zero runtime dependencies
Applied machine learning
02A compact named-entity recognizer for detecting personal information in speech transcripts, including spoken numbers and noisy, code-mixed Indian English.
12.34 ms p50 · 15.67 ms p95 · 50 runs
Document AI
03A hybrid OCR and vision-language pipeline for extracting line items from multi-page bills, keeping page context and preventing duplicate items across pages.
Multi-page extraction · duplicate-aware results
Research engineering
04A reproducible pipeline that scores Spanish transcribed responses against prompt sentences with a meaning-based 0–4 rubric, alongside calibration outputs and feature traces.
0–4 rubric · auditable scoring outputs
A little context
I build machine learning tools and AI products that need to work beyond a demo. I care about what a model gets wrong, how quickly it runs, and whether people can trust the result.
I studied at IIT Bombay and now work in AI at Qure.ai. Outside work, I build developer tools and explore applied ML through independent projects.
IIT Bombay · Civil Engineering, Computer Science minor · 2022–2026
Background
A short timeline of the teams and work that shaped my path.
Qure.ai · Bengaluru, India
Siemens
Tarini Consulting
NoQs Digital
Civil Engineering · Computer Science minor
Toolkit
A focused set of tools I use to build, evaluate, and ship useful systems.
Let’s connect
I’m happy to talk about applied AI, reliable ML systems, and thoughtful product engineering.