Back to homepage
Technical Builds
Custom tools, scrapers, and automations
I write code to solve marketing bottlenecks faster. From interactive grammar engines to automated ad scrapers and copy audit tools — explore what I've built.
-
- Claude API
- Python
- Prompt Engineering
Conversion Copy Audit & Heuristic Analyzer
- What I wanted to solve
- Manual copy auditing takes hours and often misses subtle cognitive friction points, weak objections handling, and clarity leaks.
- What I tried
- Built a structured LLM pipeline running 14 cognitive evaluation lenses (Tension, Friction, Call to Action, Specificity, Credibility) against ad copy and landing pages before spend begins.
- What happened
- Surfaced vague value propositions and weak call-to-actions before publishing, cutting copy iteration and review cycles by over 60%.
- What I learned
- AI creates far more enterprise value as an adversarial auditor and stress-tester of copy than as an unguided generator.
-
- React
- Vite
- Tailwind CSS
- Interactive UI
Ajrum: Interactive Arabic Grammar Study Engine
- What I wanted to solve
- Classical Arabic grammar texts are notoriously dense and structured for rote memorization rather than active interactive comprehension.
- What I tried
- Engineered a responsive single-page web application bundling the complete text of Al-Ajrumiyyah with dynamic color-coded syntactic breakdown, searchable charts, and interactive quiz engines.
- What happened
- Shipped as a blazing-fast, self-contained single-page web app with instant search, interactive rule cards, and mobile-friendly test modules.
- What I learned
- Complex rule-based systems become 10x more intuitive when structured visually through component hierarchies and real-time interactive feedback.
-
- React
- TypeScript
- Tailwind CSS
- Playwright
- Simulation Engine
AdsSim: Meta & Multi-Platform Ad Manager Simulator
- What I wanted to solve
- Testing paid campaign hypotheses and training media buyers on live ad accounts carries real budget risk, slow reporting feedback loops, and expensive trial-and-error.
- What I tried
- Architected an interactive, full-fidelity digital advertising simulator replicating Meta Ads Manager console chrome, Advantage+ audience configuration, CAPI event tracking, live FB/IG feed & Reels previews, and Campaign Doctor with 18 diagnostic troubleshooting puzzles.
- What happened
- Engineered a production-grade bilingual (EN/AR) simulation suite with zero financial risk, allowing rapid scenario testing, creative validation, and diagnostic campaign drills.
- What I learned
- Interactive simulation bridges the gap between marketing theory and execution, developing deep campaign intuition before committing client ad spend.
-
- Python
- Playwright
- LLM Extraction
Automated Meta Ad Library Intelligence Scraper
- What I wanted to solve
- Tracking competitor ad moves across Meta Ad Library manually is slow, scattered, and nearly impossible to synthesize at scale.
- What I tried
- Engineered an automated crawler monitoring competitor accounts, extracting live copy hooks, and clustering their messaging angles automatically via LLM embeddings.
- What happened
- Produces structured weekly intelligence reports showing active angles, creative fatigue, and gaps in competitor offerings.
- What I learned
- Turning public market activity into structured empirical data gives media buying strategy a decisive edge before a single dollar is spent.
Have a project or want to collaborate?
Email me