This role is with one of Dex’s trusted partner companies. We work closely with their teams to truly understand their culture, goals, and what they’re looking for, so we can match you with the right opportunity and give you context about the role before you commit to a process.
The role
Frontier labs do not lack compute; they lack high-signal data and rich, resettable digital environments where agents can learn complex tasks through trial and error. Backed by prominent seed investors and led by a repeat founder with a prior exit, this European AI lab scaled to more than 50 staff across London and Paris in roughly a year, building the foundational environment and data layer currently relied on by leading frontier AI labs.
As a Member of Technical Staff, you will take ownership of ambiguous research-engineering problems across data curation, model fine-tuning, and agent execution harnesses. This is not an application-layer role stitching third-party APIs together, nor is it an academic seat divorced from production; you will apply methods like SFT and RL, measure what holds up under rigorous evaluation, and turn viable prototypes into observable, hardened systems.
The work
- Design, curate, and scale training-data and synthetic-data pipelines purpose-built for agent training across complex digital workflows.
- Fine-tune and post-train models using supervised fine-tuning and reinforcement learning methods, establishing rigorous feedback loops and evaluation suites.
- Develop and harden agent harnesses and sandboxed execution environments that reliably connect models to real, interactive software tasks.
- Translate open-ended, ambiguous AI questions into technical plans and production-ready systems alongside software and curriculum engineering teams.
- Drive technical decisions and coordinate complex engineering efforts across the team without relying on line-management hierarchy.
What You Bring
- Direct, hands-on experience fine-tuning or post-training models using methods like SFT or RL, with the ability to detail precisely what was trained, how it was evaluated, and what changed.
- Deep technical grounding in AI and ML fundamentals beyond application-layer orchestration, specifically around model optimisation, RLHF, and rigorous evaluation design.
- Strong Python foundations and software engineering discipline, writing clean, testable, production-grade code rather than isolated research prototypes.
- Proven ability to break down ambiguous research-engineering questions from first principles and drive complex technical initiatives over multi-month horizons.
- Based in or willing to work hybrid from London or Paris.
#J-18808-Ljbffr…