Mehdi Amari

I worked as a Full-Stack Engineer at Tornads, building an AI-powered video ads platform across frontend, backend, and asynchronous job orchestration. On the frontend, I developed multi-step creation flows in Next.js and React for UGC ads, SaaS/app video ads, and AI video tools, including media uploads, product URL extraction, avatar and voice selection, validation, credit-based actions, and live processing states. On the backend, I built NestJS APIs, services, and orchestrators that transformed user inputs into finished video assets by coordinating LLM script generation, text-to-speech, transcription, AI image/video generation, storage, and Remotion-based rendering. I also designed and maintained the BullMQ + Redis job architecture used to run long-running workflows reliably, with dedicated queues and workers, job deduplication, retries, stalled-job handling, progress tracking, error recovery, and operational observability.

At qardapi I worked as a Senior Fullstack Engineer on AI and data-driven product development. My work involved building backend services and automation workflows with NestJS and TypeScript, integrating LLM-based features, and working with both PostgreSQL and Neo4j for structured and graph-based data management.

I also contributed to user-facing products, including interactive dashboards and conversational interfaces, with additional work around LangGraph, data orchestration, and scalable application architecture.

Designed and implemented a suite of Hardhat-tested Solidity smart contracts Architected the stateful transaction orchestration service to manage the entire lifecycle of off-chain to on-chain interactions, handling job queueing, optimistic retries, and nonce management. Implemented a comprehensive observability stack (Prometheus, Grafana) to monitor the transaction orchestration service, tracking job failures, API latency, and on-chain gas volatility