- CS Undergraduate @ Arizona State University
- SWE @ ASU Decision Theatre | Research Engineer @ RISE Lab
- Director @ ASU AI Society | Officer @ Software Developers Association
- Experienced in building Internal Software Systems, Agentic Systems, and ML pipelines
- Actively research and build in LLM, RAG & RL architectures
- Memboard — Protected DB & Utility MCP Cloud for powering Multi AI worker agents with real time dashboard (AIS)
- Bijou — Playing around with LLM architectures to spark something
- Piramid — Rust based Hybrid vector database for LLM agents with MCP support
- Kaelum — Neural Reasoning Framework for Mixture of Expert LLM agents with Monte Carlo Tree Search
- AI Pedia — Headless CMS for managing AIS Wikis and resources (AIS)
- Godfather — Admin dashboard for monitoring realtime gpu powered runpod services with custom CLI (AIS Internal)
- SparkyAI — Swarm of CRAG AI agents assisting students (ASU-Supported)
- Agents & RAG from Scratch — Foundations of agentic collaboration and retrieval architectures (AIS Workshop)
- BenchMind — AI agent benchmarking evaluation + visualization (CalHacks S12 project)
- RLHF from Scratch — PPO-based feedback optimization for LLMs (AIS Workshop)
- LLM from Scratch — Transformer-based LLM architectures built from fundamentals (AIS Workshop)
- Decoy — RAG-powered cyber intelligence pipeline (2x Google Hackathon Winner)
- SportSitters — Fullstack Chat, Voice/Video call, Booking Management iOS/android App with Admin CSM realtime Dashboard
- Platform — application integration for notion and google calendar API with faster lighthouse performance (SoDA)
- Eve — Gemini-powered robotic arm for multimodal equipment handling (PI Automation)
- Lux — Agentic AI for real-time fashion recommendations (Devhacks)
- AI Solar Panel — YOLOv8-based adaptive solar tracking system
- Carbon Emission Detector — Real-time CO₂ estimation from vehicle detection wiht TFlite phone inference
- Educational Website — Headless LMS for managing course content and payment gateways (CSYA)
- RL-Maze — Reinforcement Learning Agent for simulating maze
- Risk-Sensitive Forward Modeling for Delayed Teleoperation Using Cumulative Prospect Theory - CPT-based VAE model blending human risk perception for teleoperation safety under communication delay (inprog)
- Google Gemini SDK — Added asynchronous functions & system-level optimizations adopted in 150+ enterprise deployments.
- Intel Retail AI Suite — Developed LiDAR-based theft-detection module using OpenCV & Roboflow, reducing shrinkage by 25% in pilot tests.
- 2nd Place — Google Cyber 360 Spark Challenge '25
- 1st Place — Zoom Campus Impact Spark Challenge '25
- 1st Place — Intel HackTheVision '25 Hackathon
- 1st Place — International Codeday '24 Hackathon
- 1st Place — TechXcelerate '23 Hackathon
- 2nd Place — International Happy Hacks 1 '23 Hackathon