Working on a low-latency, multi-agent voice patient simulator with React.js, FastAPI, and Docker, allowing doctors to practice diagnosis and get treatment advice. Designed an AI-powered context retrieval pipeline (LlamaIndex + ChromaDB) and tuned prompts to mimic realistic patient responses. Achieved under 200 ms response time and shortened practitioner training by 10%.

Designed and implemented a gamified platform for skill learning, providing structured roadmaps, walkthroughs, and notes generated by fine-tuned LLMs. Integrated resources from roadmap.sh and ensured accuracy through continuous feedback loops. Achieves a 10% increase in user learning efficiency.

As part of my academic and engineering work, I designed and developed an IoT-based agricultural surveillance and automation system using Raspberry Pi and machine learning. I built a complete solution that integrates real-time animal detection (using a YOLO Tiny model), environmental monitoring with multiple sensors (for soil moisture, temperature, humidity, and pressure), and automated irrigation based on sensor data. The platform sends instant alerts to farmers via Telegram and SMS when animals are detected and provides a cloud-based dashboard for continuous visualization of environmental data. My project leverages affordable hardware and open-source tools to deliver a practical, remotely accessible farm management solution that addresses major challenges in modern agriculture.