About Me
Data analyst turned AIML engineer based in Chennai. I iterate quickly, keep code clean, and bias toward measurable, production-ready solutions.
AIML Engineer · Final-year CSE
Chandan Malakar ships OCR-to-RAG copilots ready for launch.
Curious technologist and final-year Computer Science student and unapologetic tinkerer — I turn messy data into reliable, Dockerized assistants. I specialize in building smart, scalable systems — from ML pipelines to LLM-backed chatbots — that don’t just run, they deliver. With hands-on project deployments, hackathon wins, and a love for clean code, I’m ready to bring AI from whiteboards to production.
Current focus
FinSight OCR + RAG lab
Pairing Qwen3 VL OCR, LangChain, and guardrailed LLM flows to make financial statements searchable in seconds.
- 92% extraction accuracy
- Dockerized triage APIs
- Latency < 1.2s
What I Do
- Data workflow automation
- LLM + RAG orchestration
- ML model engineering
- DevOps & CI/CD
Skills & Tools
- Python
- SQL
- C++
- Pandas
- NumPy
- LangChain
- Hugging Face
- ChromaDB
- Docker
- GitHub Actions
- AWS
- Power BI
Work
Systems engineered with intent.
May 2025 – Jul 2025 · Chennai
UNIDIF CORP · Data Analyst Intern
Embedded with the manufacturing intelligence team to turn raw production data into decisions.
- Automated Python + SQL pipelines for 1k+ records, eliminating 40% manual prep.
- Ran regression + hypothesis tests that exposed defect trends and improved quality KPIs.
- Published weekly insight packs and partnered cross-functionally on data-backed actions.
On campus · 2024 – Present
AIML builder & student leader
Lead end-to-end OCR/RAG builds, mentor peers, and represent GenAI programs on campus.
- Google Student Ambassador (Gemini Edition) and Netrix Ideathon winner.
- Runner-up at Intellecthon 2025 for shipping an OCR insights copilot.
- Created repeatable templates for Docker, CI/CD, and demo storytelling.
Projects
OCR + RAG builds that prove the point.
Nov 2025 – Dec 2025
FinSight · OCR + LLM bank chatbot
Qwen3:VL-OCR → LangChain → Hugging Face pipeline that cleans messy statements and serves realtime answers.
- 92% extraction accuracy across multi-format PDFs.
- Guardrailed RAG flows with verifiable responses.
- Dockerized delivery for rapid QA.
Oct 2025 – Nov 2025
Lexora · PDF AI assistant
Embedding-powered PDF query engine with vector search and source-mapped answers for enterprise ingestion.
- ChromaDB store with custom embeddings.
- Upload, query, and status APIs for integrations.
- Monitoring hooks for latency and drift.
Jul 2025 – Aug 2025
LegalX · Legal advisory RAG system
Retrieval engine over 10k+ case files with transparent citations for compliance teams.
- Transformer retrieval plus citation mapping.
- LangChain orchestration with guardrails.
- Prompt telemetry for trustable QA.
Resume
Download the full story.
Two pages covering education, sprint-by-sprint experience, certifications, and awards.
Contact