About the Client

We're working with an independent Artificial General Intelligence Lab that is pushing the limits of agentic AI, structured reasoning, and compact model architectures. They’re a small but well-funded team (around $5M ARR) operating right at the edge of what’s possible. Their focus isn’t on simply scaling bigger models, but on building systems that can reason, act, and persist over time.

They don’t care about degrees, titles or pedigree — what matters is capability. If you can reason deeply, move fast and push technical boundaries, you’ll thrive there.

The Role

They’re looking for an Applied Scientist / Research Engineer to take ownership of a research track and drive it forward. Depending on your expertise, this could mean working on compact model design, reasoning benchmarks, or agent-specific training strategies.

The role is highly independent — you’ll be trusted to explore bold ideas, design and run experiments, and turn insights into working systems. You’ll also collaborate with systems engineers to scale your research into production.

Key Responsibilities

• Design and run end-to-end experiments, from hypothesis to evaluation
• Build and refine reasoning-focused evaluation suites
• Develop and test new training strategies for agentic systems
• Partner with systems engineers to integrate and scale outputs
• Stay on top of the latest AI/ML research and contribute original ideas

Requirements

• Strong applied ML research experience (reasoning, compact models, or agent systems ideal)
• Solid grounding in maths, experimental design and statistical analysis
• Proven ability to independently lead and deliver research projects
• Comfort moving quickly from idea to implementation in a fast-paced, exploratory environment

Why This Role

This is a chance to join a small, well-funded team where your work will have direct impact. You’ll have the freedom to pursue meaningful research, access to resources, and the opportunity to shape the future of agentic AI.

Salary: $200K-300K

E: [email protected]