Research Scientist Intern - PhD(AI Agentic Systems)

Cornerstone VC; Greater London; Hybrid; GBP 17,000 - 28,000; Full time

Benefits offered by this job

Remote-friendly setup

Office in London

Option to extend to 6 months

Conference travel support

Job summary

Leeroo—a Y Combinator-backed company—seeks Research Scientist Interns to advance its Learning Engine with continuous-learning agents. You’ll design, implement, and evaluate new algorithms, publish results, and move prototypes into production, directly impacting real users.

You’ll collaborate with founders and senior researchers, benefiting from generous cloud compute and opportunities to travel for conferences and open-source work.

Qualifications

• PhD in AI, CS, or equivalent research excellence.
• Publications at top AI conferences or an open-source project with significant traction.
• Strong Python and PyTorch experience.
• Excellent problem-solving ability; able to work independently and in teams.
• Familiarity with graph databases, vector stores, or Elasticsearch for data retrieval.
• Knowledge of advanced training algorithms (DPO, RLHF, GRPO).
• Excellent written and verbal communication.

Responsibilities

• Design and run state-of-the-art ML experiments to extend Learning Engine.
• Publish results through papers, datasets, and open-source code.
• Collaborate with product and engineering to translate research into the Learning Engine, delivering value to customers across diverse domains.

Skills

Python

Problem solving

Communication

Research collaboration

Publications awareness

ML fundamentals

Education

PhD in AI/CS or equivalent

Tools

PyTorch

Job description

Location: Remote (EU, UK, US) / In‑person in London

Duration: 3 months (option to extend to 6)

At Leeroo—a Y Combinator‑backed company founded by former Meta AI and LinkedIn AI researchers and engineers, we’re building continuous‑learning agents driven by our Learning Engine, a modular framework that lets agents learn from human knowledge bases, feedback, and through self‑discovery, then apply that learning—evolving into expert teammates that get better every day rather than demo‑stage agents. We’re seeking Research Scientist Interns to join us in advancing the Learning Engine to build superhuman agents across domains.

About The Role

You’ll partner directly with the founders to design, implement, and evaluate new algorithms that let agents learn continuously. Successful prototypes move quickly into production, so your research will have an immediate impact on real users.

Responsibilities

• Design and run state‑of‑the‑art machine‑learning experiments to extend our agentic Learning Engine.
• Publish results through papers, datasets, and open‑source code.
• Collaborate with product and engineering to translate your research findings into the Learning Engine, delivering value to customers across diverse domains.

Qualifications

• PhD in AI, Computer Science, or a related field — or an equivalent record of research excellence
• Publications at top AI conferences (Neurips, ICML, ACL, …) or an open‑source project with significant traction
• Strong Python skills and experience with ML frameworks such as PyTorch
• Exceptional problem‑solving ability; comfortable working independently and in teams
• Familiarity with graph databases, vector stores, or Elasticsearch for structured‑data retrieval
• Knowledge of advanced training algorithms (e.g., DPO, RLHF, GRPO)
• Clear written and verbal communication

What We Offer

• Remote‑friendly setup — work from anywhere in compatible time zones, with our office based in London for in‑person collaboration
• Competitive, top‑of‑market compensation package
• Freedom to pursue ambitious research aligned with our mission
• Close collaboration with the founders and senior researchers
• Limitless cloud compute — GPU clusters and a war chest of cloud credits ready for your experiments
• Conference travel, publication, and open‑source support
• Option to extend to 6 months or transition to a full‑time role

If you’re excited about creating AI that learns continuously and autonomously, we’d love to hear from you.