Skip to content
CHIRAG_VIJAY
[INIT_SEQUENCE]
VOL.01 // 2026

I make machines that read, think, and (mostly) tell the truth.

I build AI systems that have to survive real users, not just a demo — retrieval that cites its sources (Lexora) and vision models that turn messy documents into clean, structured data (DocLayout-YOLO).

SECTION_01

Selected Works

[ENTRY_01][LEX-01]

Lexora

RAG-based legal research assistant for Indian law.

V1 live
[LIVE DEMO]
[ENTRY_02][DOC-01]

DocLayout-YOLO

Fine-tuned YOLOv8 that detects a document page's semantic layout and extracts its text.

MIT · v1.0.0
[LIVE DEMO][GITHUB]
SECTION_02

Log Archive

I Built a Production RAG System for Indian Law — and Refused to Trust It Until I Measured It

THow I turned a stack of legal PDFs into a chatbot that cites the exact section of the law — and why the evaluation scores, not the demo, were the real project.

[AUG_2026]
SECTION_03

About [CHIRAG]

I build AI systems that have to survive real users, not just a demo. Lately that means retrieval pipelines that cite their sources (Lexora) and vision models that turn messy documents into clean, structured data (DocLayout-YOLO). I like the unglamorous parts — evals, latency, the numbers that decide whether a model actually works. Currently an AI/ML Engineer at TCS iON.

CORE STACK
  • -Python
  • -LangChain
  • -FastAPI
  • -Azure OpenAI
  • -ChromaDB
  • -Docker
Available for full-time roles.
FIG. 1 // NODE GRAPH