Projects

Power and timing optimization using multibit flip-flop banking with Greedy, K-Means, and Mean-Shift clustering algorithms.

Implemented optimization passes including algebraic identity simplification and constant folding, achieving 15% instruction reduction.

Embedded system integrating STM32 firmware and Python-based host utilities for real-time audio processing and control.

Sobel edge detection kernel module on Raspberry Pi with ARMv8 NEON SIMD optimization and PMU performance analysis.

Trained ensemble models (Logistic Regression, Random Forest, LightGBM, CatBoost) for home team outcome prediction.

Hybrid quantum-classical framework integrating qGAN and Quantum Signal Processing for financial risk analysis.

CORDIC-based QR decomposition accelerator for 3×3 symmetric matrices with RTL to gate-level verification on 16nm library.

Developed a parallel memory-trace analyzer for detecting false sharing in multithreaded programs. Compared sequential, mutex-based, sharded, and thread-local reduction approaches, and evaluated their scalability and throughput using synthetic traces and real benchmarks.

Validated across multiple full-chip designs with 10K+ violations, achieving 0 unwaived errors while reducing full-chip repair time by over 10× versus manual repair.