Computer Science Undergraduate | Systems Engineering & AI Research Specializing in Computer Vision, Network Forensics, and Apple Silicon Optimization

Stack: Meta SAM 2, OpenCV, SQLite, Python Developed a hybrid vision pipeline to track object "topology changes" (fragmentation) where standard foundation models fail.

• Innovation: Engineered a "Multi-Centroid Initialization" strategy and Morphological Erosion logic to maintain Parent ID persistence during object fracture.
• Implementation: Integrated an SQL telemetry backend for automated incident logging and audit trails. View Repository

Stack: Scapy, Scikit-Learn (Random Forest), macOS ARM64 (M4 Pro) A real-time Network Intrusion Detection System optimized for high-performance inference on Apple Silicon.

• Technical Logic: Bridges low-level packet ingestion (Promiscuous Mode) with ML inference to classify traffic as "Normal" vs "Attack."
• Performance: Achieved sub-millisecond classification times using asynchronous sniffing and localized virtual environments. View Repository

Stack: Scikit-Learn, Pandas, Time-Series Analysis A predictive model forecasting pollutant levels in Lucknow, India, to provide data-driven public health recommendations.

• Methodology: Engineered lag and rolling features to capture temporal patterns in real regional air quality monitoring data collected by CPCB (Central Pollution Control Board).
• Impact: Built a classification pipeline to categorize AQI levels and automate safety protocols. View Repository

Autonomous Quadcopter Platform (STM32) Core: STM32F103C8T6 (Cortex-M3) | C/C++ | Embedded Systems Engineered a functional quadcopter from the ground up, focusing on PID control loops and sensor fusion.

• Legacy: This project established my foundation in low-level ARM architecture and real-time hardware constraints.
• Roadmap: Currently architecting a Vision-Upgrade for this platform to integrate autonomous navigation and obstacle avoidance using my current CV research.

LinkedIn | Email