Research Scientist/Engineer - General Decision & Control Agent
SKILLS
FULL DESCRIPTION
Research Scientist/Engineer - General Decision & Control Agent
[Employer hidden — sign up to reveal] is seeking a Research Scientist/Engineer to join the General Decision & Control Agent team. The role involves designing agent memory systems, developing self-evolution capabilities, and investigating reinforcement learning optimization techniques for agents.
About [Employer hidden — sign up to reveal] Limited
Founded in 1987, [Employer hidden — sign up to reveal] is a leading global provider of information and communications technology (ICT) infrastructure and smart devices. We have 207,000 employees and operate in over 170 countries and regions, serving more than three billion people around the world. Our vision and mission is to bring digital to every person, home and organization for a fully connected, intelligent world. This spirit of innovation has led [Employer hidden — sign up to reveal] to work in close partnership with leading academic institutions in the UK. For the past 30 years we have maintained an unwavering focus, rejecting shortcuts and easy opportunities that don't align with our core business.
Job Summary
London Office of UKRD is dedicated to researching and building sustainable, generalizable and evolvable Agent systems. We believe that enabling models to operate stably over extended periods across diverse task environments with continuous self-iterative optimization is an essential path toward artificial general intelligence. We are seeking passionate, creative talents with expertise in Agents and reinforcement learning to join our team and conduct cutting-edge research together.
Key Responsibilities
- Agent Memory: Participate in the design and development of Agent memory systems to equip Agents with ultra-long context processing capabilities and address memory degradation issues in long-running execution environments.
- Agent Self-Evolution: Work on building self-evolution capabilities for Agents. Explore unified representation and optimisation techniques for Agent Harness to realise continuous evolution of the Agent Harness framework.
- Agentic RL: Investigate Agent optimisation techniques based on parametric and non-parametric reinforcement learning, and build collaborative update pipelines linking Agent policy models and Agent Harness.
Person Specification
Required:
- Solid engineering implementation skills OR profound theoretical foundations (candidates must excel in at least one of the two, with promising potential in the other).
- In-depth theoretical understanding or hands-on practical experience in related fields including Agents, LLM reasoning and reinforcement learning.
- Commitment to sufficient working hours: full-time researchers work five days per week; interns are required to work 2–3 days per week for a minimum duration of six months.
Desired:
- First-author publications at top-tier conferences or journals including ICLR, NeurIPS, ICML, ACL, EMNLP and so on.
- High-citation research papers or influential open-source Agent projects.
- Outstanding rankings in relevant competitions such as Kaggle and ARC-AGI.
Location: [Employer hidden — sign up to reveal]