Thomas P.

AI automation and AI evaluation. I design and run automation systems for clients: LLM pipelines, agents, API and CRM integrations, with n8n self-hosted on Docker, the OpenAI and Anthropic APIs, Supabase, Next.js, Telnyx, Typebot and HubSpot, including the server administration. My code and architecture documentation are public on GitHub — the link is in my portfolio below. I also work on AI evaluation: LLM output and response quality, pairwise preference data, instruction following, speech and TTS assessment, French pronunciation and prosody, EN→ES dubbing QA, and annotation against detailed guidelines. French (native), English and Spanish (professional).