Ahmad Madi

Senior Software Engineer specializing in backend systems and applied AI on AWS. Since 2019 I've built the internal platform a logistics company runs on daily — event-driven automation, operational analytics, and the reliability work that keeps it up. Over the past year I've put LLMs into that production system: a transcription-and-summarization pipeline covering every support conversation, and an agent harness that carried a 56-screen UI migration.

Same discipline either way: make it reliable, keep it cheap, don't break what people already use.

Senior Software Engineer at Washmen (2019 – Present)

• Built and shipped an AI pipeline that summarizes ~6,000 support calls and several thousand chats per month for ~$1.60/day in inference cost: AssemblyAI transcribes Ziwo call recordings, Amazon Nova Lite on AWS Bedrock writes the summaries, and results land on each customer's timeline in the ops dashboard with in-place call playback and multi-conversation summary merging.
• Replaced manual conversation review for the CX team — what agents used to reconstruct by hand (what was said, promised, and when) is now available at a glance; live in production and in daily use.
• Designed and built an agent-driven UI migration system (Claude Code + Playwright): a work queue, a written playbook the agent follows, and ~15 automated gates — accessibility, visual hierarchy, layout/overflow, design conformance, and scope isolation — that must pass before anything ships.
• Migrated 56 dashboard screens off four legacy UI stacks (MUI, react-bootstrap, Kendo, SCSS) to shadcn/ui + Tailwind with it, cutting per-screen effort from days to under an hour; the system feeds review lessons back into its playbook, and later screens shipped with zero escaped UI bugs. Every change is human-reviewed, additive, and revertible with a one-line route change.
• Increased the performance of a critical API by 75×, resolving ECS health failures and stabilizing real-time processing.
• Rebuilt the operations tooling into an API-driven system — migrating legacy PHP/jQuery modules to React + TypeScript — and moved the whole team across with zero downtime.
• Built event-driven automation on AWS Lambda (Node.js/Python) that removed manual steps from daily operations, plus the backend services powering operational analytics used across multiple business units.
• Architected a cloud-native audit logging system (Lambda + DynamoDB + PostgreSQL) for compliance-grade change tracking, and a near-real-time search indexing service (Lambda + Algolia).
• Mentored junior developers and contributed to cross-team decisions on API design, cloud infrastructure, and event-driven architecture.

Bachelor of Computer Science in Computer Science – Lebanese International University (2016 – 2019)