Research Engineer – Post-Training & Reasoning
Location: San Francisco Bay Area (Onsite)
Compensation: Highly Competitive + Meaningful Equity
We're partnering with one of the most ambitious AI startups in the world, backed by leading investors and building at the frontier of foundation model research.
Rather than building another chatbot or application layer, this team is focused on a far more ambitious goal: creating AI systems capable of accelerating and automating AI research itself.
They're looking for exceptional Research Engineers who want to work on the next generation of post-training techniques, reasoning models and autonomous research systems.
What You'll Work On
- Develop novel post-training algorithms for frontier language models.
- Advance reasoning capabilities through reinforcement learning and large-scale experimentation.
- Design and improve evaluation frameworks that measure genuine model capability.
- Build scalable training and experimentation infrastructure.
- Research new approaches to synthetic data generation, alignment and model improvement.
- Work alongside a small team of world-class researchers with significant ownership from day one.
We're Looking For
You'll likely have experience with several of the following:
- Reinforcement Learning (GRPO, DPO, PPO, RLHF or related methods)
- LLM post-training
- Preference optimisation
- Reward modelling
- Evaluation and benchmarking
- Synthetic data generation
- Reasoning models
- Large-scale distributed training
- PyTorch
Ideal backgrounds include researchers from organisations such as Meta Superintelligence, OpenAI, Anthropic, Google DeepMind, Microsoft AI, xAI, Mistral, Amazon AGI, NVIDIA, Poolside or other frontier AI labs.
We're particularly interested in individuals with around 2–5 years of professional experience who have already made meaningful contributions to frontier model development.
What Makes This Different?
- Opportunity to own research rather than optimise existing products.
- Small, highly technical team where every researcher has a significant impact.
- Work on genuinely novel research problems at the frontier of AI.
- Access to substantial compute resources and the freedom to rapidly test new ideas.
- Competitive compensation alongside meaningful early-stage equity.
If you've worked on post-training, reasoning, evaluation, RL or foundation model research and are interested in helping shape the next generation of AI systems, I'd be happy to share more details in confidence.
