RTCP
A full-stack, Discord-like real-time chat platform featuring a microservices backend in Go and a modern React frontend. Built with scalability, reliability, and developer experience in mind.
Mammography Analysis (Teknofest: AI in Health)
Designed a Computer Vision system for mammography images: BI-RADS classification & lesion detection (mass/calcification). Applied deep learning models (ResNet, YOLO, DE-DETR, Xception) with custom attention mechanisms and advanced loss functions.
OptiTask (Workload Management)
Built a role-based workload management system that dynamically reallocates tasks among team members. Applied Repository Design Pattern and secured the system with HTTP-Only JWT Cookie Authorization & Authentication.
CUDA - Process Threat Detection
Developed an anomaly detection system analyzing process resource usage with CUDA-based parallel computation. Built an ML pipeline for classifying processes as malicious or benign. Integrated OS-level data collection via JNI with Java, achieving real-time detection capability.
FinSight (Open Banking & Agents)
Built a multi-agent system (Portfolio, Market, Strategist, Explainability agents) with an orchestrator for real-time financial insights. Combined LLM integration and data-driven strategies for user financial queries.
ReCall
Multi-modal on-device semantic search for Android — indexes screenshots, notes, and voice into a local vector DB with hybrid BM25 + embedding search and cross-encoder re-ranking.
BeHave
On-device behavior design app running Gemma 3n (INT4) fully offline — 10–30 tok/s on mid-range Android devices, zero cloud dependency.