Senior Full-Stack Engineer — LLM evaluation, agentic workflows, advanced developer tooling Carlsbad, CA · tn76.com · LinkedIn · tounsils@gmail.com

I build production web applications end-to-end and design evaluation tasks that benchmark frontier AI agents. Recent focus: deterministic test harnesses, long-context puzzle design for LLM agents, and CLI-driven developer tooling.

• LLM evaluation tasks for an agent-benchmark program (Python, Docker, pytest). Authoring stateful environments with hidden constraints, mislabel traps, and seasonal-gate puzzle mechanics that calibrate agent difficulty against models like Claude Opus 4.6 and GPT-5.
• Timecards — a local-first desktop time tracker for managing multiple concurrent projects (repo).
• Digital QR Card — full-stack platform (Node, Express, MongoDB, Vite + React) at digitalqrcard.com.

Languages — Python, TypeScript, JavaScript, PHP, C++, Bash Frontend — Next.js, React, React Native, Vue, Tailwind, Bootstrap Backend — Node.js, Express, Laravel, REST APIs Testing — pytest (boundary + edge-case validation, fixture-driven integration), Jest, deterministic harnesses AI / LLM — agent evaluation, prompt-trap design, long-context puzzle authoring, RLHF-adjacent annotation workflows, Anthropic + OpenAI APIs DevOps / Cloud — AWS (Solutions Architect track), Docker, Kubernetes, Jenkins, Nginx, CI/CD Databases — MongoDB, PostgreSQL, MySQL, MariaDB Tooling — Claude Code, advanced CLI automation, WSL, git workflows

• Email — tounsils@gmail.com
• LinkedIn — in/mohameditounsi
• Stack Overflow — 10537019/mohamed-tounsi
• HackerRank — tounsils

_{Open to senior full-stack roles with an LLM-evaluation, agent-benchmarking, or developer-tooling component.}