Real-time hazard detection system for mines. Won 3rd in Social Impact and 4th in Hardware at Yale's YHack against 700+ participants. Features a live data pipeline integrating gas sensors and depth cameras via Raspberry Pi, plus a simulation to map tunnel geometry and model gas diffusion.

32-bit RISC-V integer microprocessor with Harvard architecture, designed in SystemVerilog. Features an FSM-based multicycle control unit, full RV32I instruction set support, and hardware peripheral I/O. Verified with Icarus Verilog and deployed on an iCE40UP5K FPGA.

Fully automated cleaning station combining AI-powered dirt detection, rotary encoder-based disc handling, and servo-actuated brush control. Built as a complete mechatronic system for Olin's PIE course, featuring a Python state machine, Roboflow CV model, and real-time LCD feedback.

Inverted pendulum balancing robot built on a Pololu Balboa platform. Designed a 5-pole PI controller for simultaneous angle stabilization and stationary position hold, modeled in MATLAB/Simulink, then empirically tuned on hardware to handle real-world motor delay and saturation.

Bio-inspired, weed-uprooting robot for rice fields. Won 1st Place and $5,000 in seed funding at the Babson College Buildathon out of 500+ participants. Uses a trained Computer Vision model to differentiate invasive weeds from crops, with Arduino-driven motor control for targeted uprooting.

A fully playable tower defense game built from scratch in Python using Pygame. Features multiple tower and balloon types with strategic gameplay mechanics.

A desktop application that maps real-time hand gestures from webcam input to emojis using computer vision and machine learning.

Designed and built an autonomous robot to navigate a taped track using closed-loop feedback control with sensor integration.