Pune, India · Available for the next hard problem
Computer Engineering student building agentic systems and realtime products: software that listens, reasons, uses tools and gets a real task finished.
01About
Hi, I'm Uddhav.
I'm a B.Tech Computer Engineering student in Pune who got hooked on AI the moment a model did something I didn't expect. Since then I've been building: realtime classroom tools, learning platforms and small ML projects, and writing about what breaks along the way.
My goal is to build AI agents that are reliable enough to trust with real work, and to grow into a team that ships them at scale.
- 90+
- Google Skill Badges
- 4
- Certifications
- 2
- Internships
02Experience
Backend AI Engineering Intern (Virtual)
Jul 2026 – Aug 2026FlyRank AI
Built a FastAPI REST API with full CRUD, input validation, correct HTTP status codes and auto-generated OpenAPI/Swagger docs; published to GitHub with staged commits and a README.
PythonFastAPIOpenAPIGCPMicrosoft Learn Student Ambassador
Jan 2026 – PresentMicrosoft
Delivered technical sessions on Python, AI and developer tools.
PythonAIDeveloper ToolsOpen-Source Contributor
Jan 2026 – Apr 2026PrivGPT Studio
Contributed to a privacy-first AI workspace with local and cloud LLMs and offline fallback; implemented multimodal cross-referencing across PDFs, video, images and audio (with transcription).
TypeScriptPythonNext.jsFlaskMongoDBAI-ML Virtual Intern
Apr 2025 – Jun 2025EduSkills
Built an ML pipeline for treatment and diet recommendations on a real-world dataset; integrated Firebase auth, Twilio and Cloudinary with 98% service reliability.
PythonTensorFlowFlaskFirebase
03Tools & Projects

Recover.ai: Agentic Payment Recovery
AI agent that detects failed Razorpay subscription payments through signature-validated, idempotent webhooks and runs a Diagnose → Decide → Guardrail → Act pipeline. A deterministic guardrail stops ambiguous LLM diagnoses from bypassing policy; every action is written to a PostgreSQL audit trail first. In a simulated 60-case batch it recovered 14 cases (23.3%, Rs 10,886) and left 46 for human review.

Autonomous Coding Agent
Research-driven AI software-engineering agent that works on existing repos in a controlled workspace, planned across 15 phases (context selection, tool permissions, verification, evaluation). Secure workspace layer rejects path traversal and symlink escapes and uses atomic writes; covered by unit and integration tests.

Bharat-Setu
Zero-cost, pilot-ready platform with satellite-verified carbon credits, a village digital twin and crisis detection, replacing 8 AWS services with open-source equivalents. Fraud protocol auto-approves under 10% variance and freezes accounts over 30%; 76 unit tests, React admin dashboard and a voice-enabled PWA.
From the Blog
Latest Thoughts
04Now / Focus
What I'm building toward.
AI agents that take real actions, realtime AI interfaces, and the dependable infrastructure underneath them.
05Contact
Let's Connect
Open to internships, collaborations and product builds. Pick a topic and tell me what you are working on.
Open to opportunities
uddhavbhople5@gmail.com