Mohammed Madi

I've built production AI systems at CybWok, including a LangGraph-based multi-agent clinical triage platform combining deterministic rules, LLM patient interviews, voice triage, RAG retrieval, and FHIR-compliant records. I also developed provenance tooling with DAG reasoning traces and deterministic replay, plus a prompt-evolution engine that reduced token usage by 74.5% while maintaining 96.7% accuracy. At Jawaker, I integrated Mintroute voucher redemption for more than 10 million registered users and rewrote the Android payment flow. At EastNets Holding, I created synthetic banking data for anti-money-laundering training, built a graph neural network for money-laundering detection, and delivered an explainable ML dashboard. I've also shipped full-stack features with .NET, Vue.js, TypeScript, and PostgreSQL at Miaplaza. My projects span game-theoretic LLM reasoning, PDF malware detection, cybersecurity challenges, and plant-disease classification. I'm ranked third globally in Huawei ICT Innovation 2024, and I enjoy sharing what I learn through cryptography, competitive programming, Python automation, and neural-network workshops.