Work / current record

Build first. Explain what changed.

A selected record of projects and experiments. Each item is labeled by its current state; detailed case studies will appear when the evidence is ready.

01

AI platform · building

PromptLens

A self-hostable TypeScript platform that captures prompts through authorized connectors, analyzes them asynchronously, and keeps recommendations and improved versions in isolated workspaces.

  • TypeScript
  • PostgreSQL
  • Redis
  • AI systems
Open repository ↗
02

Edge AI system · building

GuardWatch

A local computer-vision system for detecting drowsiness and inactivity in monitored environments, with separate development and Jetson-oriented pipelines.

  • Computer vision
  • Python
  • Edge AI
Open repository ↗
03

Offline-first app · building

Context OS

An Android context system that helps recover where work stopped across areas, projects, tasks, sessions, and inbox items without relying on a backend.

  • Flutter
  • Dart
  • SQLite
04

Data dashboard · documented

YouTube Analytics Dashboard

A Streamlit dashboard that turns exported YouTube Analytics data into channel insights, video comparisons, engagement metrics, and percentile benchmarks.

  • Python
  • Streamlit
  • Analytics
Open repository ↗
05

AI product · building

AION Image Generator

An Arabic-first social-media image generator that collects campaign context and routes image generation through a server-side proxy.

  • React
  • Vite
  • Generative AI
Open repository ↗
06

Experiment · building

AI-powered data analyst

An agent-based exploration of a complete data-analysis task: inspect, compare, visualize, and explain.

  • AI agents
  • Data
  • Python
07

Practical build · documented

Messy Excel → dashboard

A workflow for cleaning imperfect sales data, checking the transformation, and producing a useful dashboard.

  • Data quality
  • Power BI
  • AI
08

Evaluation experiment · building

Arabic AI model testing

A structured comparison of Arabic models across dialect understanding, data extraction, analysis quality, and hallucination.

  • LLMs
  • Arabic
  • Evaluation
09

Technical exploration · exploring

Vector database exploration

Learning how retrieval systems store and find semantic context—and where the abstraction becomes a real engineering tradeoff.

  • RAG
  • Embeddings
  • Systems

Evidence over claims.

The strongest project story is not “I used an impressive stack.” It is the problem, the decision, the failure, the result, and what the next question became.

Public repository and demo links will be attached when each artifact is intentionally approved for publication.

See what is in progress ↗