Staff Research Engineer: Post-Training RL & Sandbox Systems

Dex; Greater London; Presencial; GBP 110.000 - 170.000; Jornada completa

Descripción de la vacante

Dex is building frontier AI labs in London and Paris, focusing on post-training strategies, sandboxed environments, and robust data pipelines. You will design and implement end-to-end post-training processes, enabling agents to learn from real software feedback while ensuring safe, reproducible iterations across research and production teams.

Ideal candidates bring deep ML fundamentals, production-grade Python, and a track record in SFT/RL.

Formación

• Experience training/fine-tuning models (SFT/RL/RLHF) with data regimens and loss dynamics.
• Strong foundational ML knowledge including model optimization and RL architectures.
• Production-grade Python engineering with modular, observable systems.
• Ability to take ambiguous problems from concept to working system over multi-month horizons.

Responsabilidades

• Design and run end-to-end post-training pipelines with SFT and RL.
• Architect sandboxed execution environments and deterministic feedback loops.
• Scale training-data infrastructure from messy data to curated datasets.
• Establish evaluation frameworks to benchmark capabilities and ensure reproducibility.
• Provide technical leadership across ambiguous research directions with engineers.

Conocimientos

SFT/RL training

ML fundamentals

Python production

Ambiguity navigation

Descripción del empleo

Dex is building frontier AI labs in London and Paris, focusing on post-training strategies, sandboxed environments, and robust data pipelines. You will design and implement end-to-end post-training processes, enabling agents to learn from real software feedback while ensuring safe, reproducible iterations across research and production teams.

Ideal candidates bring deep ML fundamentals, production-grade Python, and a track record in SFT/RL.