🚀 Welcome to my GitHub Profile!

I bridge the gap between ideas and reality with Agentic AI and Full Stack Development.

🤖 Agentic Coding / AI-Assisted Development

(A modern, high-value addition to my workflow)

Leveraging Antigravity and Cursor for prompt-driven development, rapid prototyping, and dependency-aware refactoring.

✅ Tech Stack

Programming Languages

Frontend

Backend / APIs

AI / ML / Data

Databases

Cloud / DevOps / Infra

Automation & Utilities

I specialize in Agentic AI development, computer vision, and high-performance native applications. Here is what I bring to the table and why I make a great engineering partner:

• Self-Correcting Agentic Systems: I don't just prompt APIs. I design stateful agentic loops, robust validation pipelines, and data-integrity mechanisms (e.g., implementing strict data deduplication and performance tuning in Tawjihi AI).
• Privacy-First & Edge AI Execution: Experienced in running models locally on-device (macOS/iOS) using Apple Silicon optimizations, WhisperKit, and local Swift frameworks (like SpeakType and FocusFlow), ensuring zero-latency and 100% data confidentiality.
• Advanced Applied Science: Capable of translating complex academic papers and engineering formulations into production-grade systems (such as using RAFT optical flow models combined with physics-based stress formulations in the Visual Stress Detection System).
• End-to-End Execution: From database indexing and FastAPI microservices to fluid Next.js or native SwiftUI frontends, I build systems that are responsive, modular, and scalable.

Systems that reason, recommend, and self-correct.

• Tawjihi AI – AI-Powered Customer Recommendation Platform for Banking
  • Goal: Generate hyper-personalized banking recommendations based on demographics, financial history, and real-time behavioral insights.
  • Key Achievement: Built a robust API deduplication engine that eliminated 100% of recommendation anomalies and database bloat while expanding product coverage by 33%.
  • Tech Stack: FastAPI, Next.js, Python, PostgreSQL, Pydantic, Tailwind CSS
• CAPE – Comprehensive Assessment Proctoring Engine
  • Goal: Secure, real-time proctoring engine facilitating exam security and academic resource allocation.
  • Tech Stack: React, Node.js, Astro, MongoDB, SQL

Offline-first tools utilizing Apple Silicon optimization and local ML models.

• SpeakType – Fast, Offline Voice-to-Text Dictation Tool for macOS
  • Goal: Provide zero-latency, 100% offline global voice-to-text dictation directly into any text field on macOS.
  • Key Achievement: Integrated OpenAI's Whisper model locally via WhisperKit, optimizing inference on Apple Silicon (M1+) to operate with 0% network usage.
  • Tech Stack: Swift, SwiftUI, WhisperKit, AVFoundation, AppKit
• FocusFlow – ADHD-Friendly Productivity App (Digital Prosthesis)
  • Goal: Assist individuals with executive dysfunction using smart NLP text input, haptic dopamine feedback loops, and energy-based task filtering.
  • Tech Stack: SwiftUI, Swift Concurrency, iOS 18+ SDK, @Observable state management

Integrating physics-based modeling and deep computer vision.

• Visual Stress Detection System – Micro-deformation & Structural Failure Risk Engine
  • Goal: Detect sub-pixel deformations in sequences and compute structural risk under stress.
  • Key Achievement: Combined RAFT dense optical flow with physics-based formulations (Young's Modulus, Von Mises stress) to map temporal strain and predict failure probabilities using machine learning.
  • Tech Stack: PyTorch, OpenCV, Streamlit, SciPy, Python
• Intelligent Vision System – Modular Computer Vision Toolkit
  • Goal: End-to-end computer vision pipeline from enhancement to stereo depth map estimation.
  • Tech Stack: Python, OpenCV, NumPy, scikit-learn, Streamlit