Research Engineer (LLM Performance), London
New
Looking for a research engineer job in London? Choose from 20 open roles at 8 employers. Explore real salary data and skills breakdown across remote, hybrid and on-site roles.
Free. No credit card needed.
The roles on this page sit at the boundary between research and engineering, requiring candidates to both advance algorithmic ideas and build the systems that make large-scale experiments possible. Several positions centre on reinforcement learning: one at Anthropic asks you to design large-scale RL experiments and translate findings into production training recipes, while another focuses on improving the RL training infrastructure that researchers use daily. Pretraining is a separate strand, with a role that covers model architecture, data processing, and training optimisation for large language models. On the perception side, Helsing has a position building computer vision pipelines for 3D reconstruction and localisation in real-world conditions. Graphcore's posting spans efficient training, inference, world models, and conference publication. Safety-focused work appears too, including a senior role at Lila Sciences leading post-training, refusal classifiers, and automated red-teaming. Across the sample, hands-on systems work and empirical research tend to appear together rather than separately.
The skills counts reflect mentions in posting descriptions across all postings on this page, not stated requirements, and the sample is small enough that any single employer's language can shift the rankings noticeably. Python and reinforcement learning appear in the majority of postings, with PyTorch close behind. PhD and publications are mentioned in roughly half, suggesting academic background is relevant to many but not all of these roles. C++ appears in a minority, mostly where hardware-level or low-level systems work is involved.
The employers' own advertised ranges, for individual roles at different levels — not an average and not a market rate.
The strong majority of postings on this page are listed as onsite, with only a small number stating hybrid arrangements. None are listed as remote. The hybrid postings include at least one Anthropic role explicitly noting London office attendance. For the onsite majority, no further detail about specific locations within London is given in the fact pack. Candidates should check individual postings for office addresses and any flexibility that may not be captured in the work-mode label, as arrangements are not stated in detail for most roles.
Jobs checked 1 day ago. · Guide reviewed 9 September 2026.
Search by role, company, or anything a posting mentions.
