Real-time computer vision pipelines, embedded robotics algorithms, and machine learning architectures.
- Core Focus: Real-time computer vision (OpenCV, MediaPipe, YOLOv8), applied machine learning (Scikit-Learn, PyTorch), and robotics control systems.
- Engineering Standards: Deterministic test suites, rotation-invariant geometry, low-latency inference, and modular system design.
- Current Engagement: Progree Artificial Intelligence & Robotics Internship.
| Discipline | Technologies & Frameworks |
|---|---|
| Languages | Python, C++, TypeScript, JavaScript, SQL, Bash, PowerShell |
| Computer Vision & ML | OpenCV, MediaPipe, YOLOv8, PyTorch, Scikit-Learn, NumPy, Pandas |
| Robotics & Embedded | Closed-Loop PID Control, 3-DOF Kinematics (Forward/Inverse), Arduino C++, PLC Sequence Logic |
| Backend & APIs | FastAPI, RESTful Architecture, WebSockets, SQLite, Pydantic, Streamlit |
| Developer Tooling | Git, GitHub Actions, Docker, Linux, VS Code, Postman |
Real-time computer vision system that tracks both hands simultaneously, isolates 21 three-dimensional skeletal joints per hand, and computes extended finger counts across a 0 to 10 range.
- Replaces naive Y-coordinate thresholding with rotation-invariant Euclidean distance ratios from wrist origin and pinky MCP anchor points.
- Computes radial thumb abduction, selfie mirror-mode correction, and static gesture classification.
- Dual interface: high-throughput direct OpenCV desktop window (30+ FPS) and interactive Streamlit web dashboard.
- Developed for the Progree Artificial Intelligence Internship program.
Comprehensive engineering submission archive across Robotics & Automation and Artificial Intelligence specializations.
- Robotics & Automation: Closed-loop PID line follower simulation with anti-windup clamping, Arduino C++ firmware, analytical 3-DOF planar robotic arm forward and inverse kinematics solvers, and smart conveyor PLC sequence logic with edge telemetry logging.
- Artificial Intelligence: Multi-class text sentiment classification with TF-IDF vectorization and regularized logistic regression, heuristic A* and Dijkstra 2D grid pathfinding engine, and adaptive contour object detection pipeline with polygon shape approximation.
Production-grade real-time object detection and visual analytics suite powered by YOLOv8, OpenCV, and Streamlit.
- Supports live camera feeds, video files, and high-resolution images with real-time confidence tuning and class filtering.
- Includes Region of Interest (ROI) entry/exit counting, speed estimation, and automated CSV telemetry export.
Intelligent security camera and real-time surveillance platform.
- Ingests video streams, detects motion triggers, logs intrusion events with timestamps, and streams live frame analytics via WebSockets and FastAPI.
Enterprise document and invoice intelligence application.
- Automates document parsing, tabular line-item extraction, visual bounding-box inspection, and financial analytics with a FastAPI backend and React frontend.
Structured 30-day foundational and advanced Python curriculum.
- Includes daily problem sets, algorithmic implementations, unit tests, and an object-oriented capstone software application.
- Email: shayanshah.edu.pk@gmail.com
- GitHub: @Shaayan-shah
- Location: Swabi, KPK, Pakistan