Developed a high-throughput Kanade–Lucas–Tomasi (KLT) tracker featuring a pyramidal coarse‑to‑fine architecture for large-motion robustness.
Optimized GPU v3 implementation using fused kernels, shared memory for reduced latency, and CUDA streams for asynchronous pipelining.
CUDA, C++, OpenACC, Parallel Computing, Warp/Block Reductions.
Engineered a concurrent air traffic control simulation modeling real-world airport throughput and gate allocation.
Implemented sophisticated Inter-Process Communication (IPC) and synchronization primitives including semaphores, mutexes, and condition variables to prevent race conditions in runway management.
C++, Multithreading, SFML, System Design, IPC.
Created a high-integrity data verification system using C++ to process large-scale Amazon review datasets.
Leveraged recursive hashing to build a compact Merkle tree, enabling O(log n) verification of data integrity and efficient tamper detection.
C++, Cryptography (Hashing), Algorithm Design, Data Structures.
Developed a lightweight, console-based version control system in C++ to mimic core Git functionality for local project management.
Engineered a custom file-snapshotting mechanism using hashing to track changes and implemented a directed acyclic graph (DAG) structure to manage commit history and branching logic.
C++, Data Structures, File I/O, Hashing.
Developed a non-contact mobile health monitoring application that extracts physiological signals from facial video feeds.
Applied rPPG (remote photoplethysmography) techniques and skin-pixel filtering to isolate blood volume pulse (BVP) signals from ambient noise.
Deep Learning, Computer Vision, C++, Android SDK, Signal Processing.
Built and compared vanilla vs. attention-based Sequence-to-Sequence models for Neural Machine Translation.
Integrated additive/multiplicative attention mechanisms to solve the "bottleneck problem" in encoder-decoder architectures, significantly improving translation BLEU scores.
PyTorch, Python, NLP, Deep Learning.
Designed a deep learning pipeline to predict systolic and diastolic blood pressure using pulse and respiratory waveforms.
Streamlined the data pipeline to map spatio-temporal features from MTTS-CAN outputs directly to physiological pressure metrics using custom loss functions.
Python, Keras, Deep Learning, MTTS-CAN Architecture.
Researched machine health monitoring using audio signal analysis across cross-domain industrial environments.
Utilized Deep Transfer Learning to adapt models trained on laboratory-grade data to noisy, real-world industrial acoustic environments.
Deep Transfer Learning, Audio Signal Processing, Python.
A comprehensive desktop suite for metro operations, managing ticketing, fleet logistics, and route optimization.
Designed a normalized SQL schema to handle complex relationships between routes, stations, and passenger transactions while ensuring ACID compliance.
Java, Java Swing, Microsoft SQL Server, JDBC.
Designed a privacy-focused web platform for temporary, self-destructing chatrooms with unique shareable links.
Implemented automatic data expiration logic to ensure chat logs are purged from memory/storage immediately upon room closure, prioritizing user anonymity.
Web Development (HTML/CSS/JS), Backend Security, Real-time Data Handling.
Developed a robust CRUD-based application for managing trip bookings and user profiles.
Integrated an event-driven UI in WinForms with a dedicated SQL backend to manage high-volume transactional data for trip scheduling.
C#, .NET Framework, WinForms, SQL Server.
Founded and developed a ride-sharing platform providing real-time tracking and secure booking for affordable transit.
Scaled the mobile architecture to support real-time GPS polling and dynamic fare calculation based on distance and demand.
Android Development, Google Maps API, Firebase, Mobile UI/UX.
Implemented the classic game Zuma entirely in x86 Assembly.
Optimized low-level rendering by directly manipulating the video buffer and utilizing hardware interrupts for real-time keyboard input handling.
x86 Assembly, MASM, Low-level Programming.
A feature-complete C++ clone of the popular tower defense game.
Architected a modular "State" system to handle transitions between menu, gameplay, and pause screens, minimizing memory leaks via smart pointers.
C++, SFML, OOP, Game Design Patterns.
Recreated the retro arcade experience focusing on classic sprite movement and collision logic.
Developed a grid-based collision detection algorithm to manage the interactions between the player, mushrooms, and the multi-segment centipede.
C++, SFML, Sprite Animation.
A hardware-software hybrid that uses a camera on a motorized mount to track human faces in real-time.
Implemented a PID-style control loop in Python to smooth the servo movements, preventing mechanical jitter during high-speed tracking.
Python, OpenCV, Arduino, Embedded Systems.
Developed a wrist-worn device using IMU sensors to analyze movement and determine sleep quality.
Developed a threshold-based classification algorithm to distinguish between deep sleep, light sleep, and wakefulness based on 3-axis accelerometer data.
Java, Android, IMU Sensor Integration, Machine Learning (Classification).