Udit Rawat

I'm an AI engineer based in New Delhi. Most of my work lives in agentic systems, RAG, and the layer where a language model has to actually hold up in production, not just in a demo. Outside of work, I build things mostly to understand them properly. Reading about an architecture only gets me so far, so I tend to reproduce it, break it, and see where it actually fails. That's where most of the projects below came from.

Experience

Feb 2025 to Jul 2026

AI Engineer, TapHealth

AI Engineer from Apr 2025, previously AI Intern from Feb 2025, promoted after 2 months.
Jul 2024 to Oct 2024

Project Intern, Perception & Intelligence Lab, IIT Kanpur

Curated a 200K+ image dataset and optimized diffusion based inpainting models, achieving 11.72 dB PSNR and 0.41 SSIM.

Selected Projects

Tribunal

Adversarial multi-agent fact-checker. A Prosecutor and Defender argue opposite sides of a claim over retrieved evidence, a deterministic verifier strikes any quote not found verbatim in the source, and a Judge rules. 96% quote grounding on a 32-claim eval. Ships as CLI, streaming web UI, and MCP server.
PythonLangGraphMCP

Animatic

Spec-driven engine for AI-narrated motion-graphic reels. JSON spec → Gemini TTS → GSAP composition → MP4. Live demos at @ada.max24.
PythonGSAP

glassbox

A GPT built from first principles in pure PyTorch, no HuggingFace. Attention, RMSNorm, RoPE, SwiGLU, grouped-query attention and a from-scratch BPE tokenizer, each tested before the next was added. A same-seed ablation improved validation loss with 12% fewer parameters. Ships with a live attention visualizer whose architecture diagram is generated from the model config, not hand-drawn.
PythonPyTorch

RepoSignal

Due diligence for every GitHub repo. Know before you clone. Local, reproducible scoring.
JavaScript

Quest

On-device RAG for fast coding problem lookup.
PythonFastAPIDeepseekR1