B.E. Information Technology · Army Institute of Technology, Pune · Class of 2028
I build AI-backed tools and full-stack products with a focus on evidence,
usable interfaces, and honest limits.
- 🎓 Information Technology undergraduate at the Army Institute of Technology, Pune
- 🤖 Interested in evidence-grounded AI, retrieval-augmented generation, and applied machine learning
- 🧩 I enjoy turning complex workflows into clear, reviewable software
- 🔎 I care about reproducibility, testing, and making system limitations visible
- 💼 Open to software engineering and AI internship opportunities
Evidence-grounded investigation software for coordinated payment abuse and fraud rings. RiskWeave connects transactions through shared infrastructure and activity patterns, ranks suspicious candidate rings, and gives analysts an inspectable network, timeline, and evidence trail.
- Built an end-to-end investigation workflow with React, TypeScript, Python, FastAPI, scikit-learn, NetworkX, and SQLAlchemy
- Designed grounded AI explanations that cite computed facts instead of inventing evidence or changing detection decisions
- Evaluated the network-aware model across five held-out synthetic ecosystems, reaching 0.971 mean PR-AUC and 0.951 mean F1
- Added reproducible runs, checksums, persisted findings, and automated backend and frontend tests
Synthetic-trained and synthetic-evaluated research software—not a production fraud verdict system.
A guarded echocardiography workflow prototype that turns scan review into a clear sequence: quality check, dual ejection-fraction estimation, uncertainty review, longitudinal trend, and clinical action.
- Built a polished, responsive interface with React, TypeScript, and Vite
- Created deterministic scenarios for poor scan quality, estimator agreement, estimator disagreement, and longitudinal decline
- Implemented a guided product demo and a replaceable inference adapter for future model integration
- Kept simulated inference and prototype status explicit throughout the interface
Demonstration prototype only—not clinically validated and not intended for patient care.
| Area | Technologies |
|---|---|
| AI & data | Python, scikit-learn, retrieval-augmented generation, graph analysis |
| Backend | FastAPI, Node.js, Express, REST APIs, SQLAlchemy |
| Frontend | React, TypeScript, Next.js, Vite, Tailwind CSS |
| Quality | pytest, Playwright, Vitest, Ruff, TypeScript |
| Data | SQL, relational databases, synthetic-data evaluation |
Third-party cards may load slowly or display incomplete public activity.
I'm interested in software engineering and AI internships where I can build, test, and improve real products. Reach me on LinkedIn or by email.