Computer Science undergraduate at BITS Pilani, Dubai Campus, building toward a career as an AI / Software Engineer. I work across two things that increasingly overlap: classical ML/edge deployment and applied LLM engineering.
- 🔭 Currently building production-style LLM systems (RAG, structured-output agents, developer tooling)
- 📄 First author on an IEEE-submitted paper on edge-deployed deep learning (MobileNetV2, 97.33% accuracy, deployed on a Raspberry Pi 4B)
- 💼 Completed a backend engineering internship at Amaani, a Dubai travel-tech startup (Python/Flask, REST APIs)
- 🌱 Currently strengthening: LLM agent architectures, Google Cloud / Vertex AI
- 📫 Reach me: uk4320930@gmail.com · LinkedIn
| Project | What it does | Try it | Source |
|---|---|---|---|
| Flowboard | Connect blocks into a workflow, run it, and inspect the data at every step. | Live demo | GitHub |
| Sayso | A demo airline service desk: one request becomes interactive flight, seat, and bag controls. | Live demo | GitHub |
| Hindsight | Explore agent runs on a timeline, separating API calls, model calls, and time waiting for a person. | Live demo | GitHub |
Start with Flowboard: open the demo and press Run. No installation, account, or API key is needed for the built-in example.
If a project is useful to you, consider giving its repository a star. Feedback and bug reports are welcome too.
Document Q&A Agent (RAG) Retrieval-augmented generation service — chunks and embeds documents with the Gemini API, indexes them in FAISS, and serves a Flask REST API that returns grounded, source-cited answers.
Support Ticket Triage Agent LLM agent that classifies customer support tickets and drafts replies as schema-validated structured JSON, using Gemini's structured output mode + Pydantic.
PR Diff Summarizer
Developer-tooling agent that turns a raw git diff into a structured PR summary — risk level, key changes, and testing notes — instead of a human writing it by hand.
Languages: 
AI / ML: 
Backend: 
Tools: 
