Research Scientist

Harrison Clarke; San Francisco (CA); Hybrid; USD 180,000 - 240,000; Full time

Job summary

Harrison Clarke in the Bay Area seeks an AI researcher/developer focused on applied research that ships into product. You will work close to code, turning research into tangible improvements in our frontier technology.

You'll join a small, fast-moving team with ownership and ambiguity, focusing on agentic systems, post-training, continual learning, and evaluation. Publications are a plus, but shipped impact matters most.

Qualifications

• Applied research that translates to shipped product improvements.
• Strong engineering ability and practical coding skills.
• Experience with AI/ML systems and evaluation design optional.
• Comfort with ambiguity in a lean, fast-moving team.

Responsibilities

• Translate research into shipped product features and improvements.
• Own parts of the system and contribute to codebase.
• Collaborate across research, engineering, and product to ship impact.

Skills

Applied research

Engineering ability

AI-native builder

Code proximity

Job description

We're working with a well-backed AI startup building frontier technology in a high-impact, real-world domain. The team is small, technical, and moving fast, with research sitting right at the core of the product.

This is applied research - your work ships into the product and directly improves how the system performs. If you like doing real research and seeing it in users' hands quickly, rather than working purely on benchmarks or papers, this is that kind of environment.

Areas of focus (depending on your background):

• Long-horizon and agentic systems - planning, tool use, and reliable multi-step autonomy
• Post-training - SFT, RL, preference optimisation, and evaluation design
• Continual learning and adaptation - how systems learn, retain, and improve over time without constant retraining

What we're looking for:

• A strong applied-research track record where your work has translated into shipped improvements
• Depth in one or more of: agentic systems, post-training, continual/lifelong learning, memory and adaptation, or evaluation
• Genuinely strong engineering ability - you build, not just publish
• Comfortable with high ownership and ambiguity in a lean, fast-moving team
• An AI-native builder who's excited to work close to the code

Details:

• Location: Bay Area - onsite/hybrid
• Comp: Competitive base + meaningful equity
• Process: Fast and lightweight - a matter of days, not weeks

Publications at top venues are a plus, but shipped, applied impact matters more.