Research Engineer - Code Generation

$250,000 – $400,000 + Equity

Bay Area (onsite)

The Company

This is a rare opportunity to join a well-funded frontier AI research lab at seed stage, backed by tier-1 investors and strategic partners at the forefront of AI infrastructure.

The founding team brings deep research pedigree with alumni from leading AI organizations including NVIDIA and Meta, and the lab has attracted backing from some of the most respected names in AI research globally.

The company is operating at the cutting edge of spatial intelligence and interactive world building - one of the most technically ambitious and least-solved problem spaces in AI today.

The Role

As a Research Engineer focused on Code Generation and Reasoning Systems, you will sit at the intersection of language models, agentic systems, and interactive environment construction. You'll be working on foundational problems with high ownership and direct research-to-product impact.

Responsibilities include:

• Designing and building agent systems that generate structured logic and interactive world content
• Developing tool-using reasoning systems for complex, multi-step content generation tasks
• Post-training and fine-tuning of code generation models, with a focus on improving reasoning, tool use, and environment construction
• Running experiments across the full stack - from model training through to real-time deployment constraints

Key Skills & Experience

• Strong experience in post-training techniques: RLHF, DPO, GRPO, or related reinforcement learning from feedback methods
• Hands-on experience with fine-tuning and supervised fine-tuning (SFT) of large language models for code generation or reasoning tasks
• Proficiency in Python and PyTorch, with deep familiarity working with transformer architectures
• Experience building or working with LLM-based agent systems and tool-use frameworks
• Solid fundamentals in ML/deep learning and software engineering
• Strong signals: competitive programming background, top-tier research publications, or early-stage founding experience

Why This Role

The problems being solved here - persistent, interactive, structured world generation - are unsolved at scale. You'll have the autonomy to own significant technical directions, contribute to foundational research, and see your work shape both the product and the broader field of spatial AI.

If you're motivated by high-impact work, deep technical ownership, and the chance to define how AI agents learn to perceive and act in the world, this is worth a conversation.

Apply

Please apply via this listing or reach out directly. All applications are handled in strict confidence. Clients will not be disclosed at initial stage.