About me

I’m Miguel Brandão, a Python-focused engineer working on agent evaluation platforms and tasks for large language models (LLMs). Recent work includes:

• ControlArena (UK Government): task/setting development with production merges across core and infra repos.
• LinuxBench (Redwood Research): platform tooling and task development from project inception (internal/private).
• METR: human baselines, task development and reliability work acknowledged in HCAST.

Previously, I was a Senior Software Developer at Softinsa–IBM, building backend systems in Python (FastAPI/Flask), Docker, and Kubernetes.

Always happy to connect—feel free to reach out via LinkedIn or email.