AI engineer for investment teams

BlueCrest Capital Management; Greater London; On-site; GBP 90,000 - 130,000; Full time

Job summary

BlueCrest Capital Management's AI Team in Front Office Technology is hiring a platform-focused AI engineer. You will build central capabilities, run production inference on GPUs, and optimize model serving across teams of PMs, traders, quants and researchers.

You will implement scalable AI services, governance, and cost-aware deployment, using tools like vLLM, LoRA, QLoRA, and Kubernetes. This role combines software engineering rigor with real-time investment workflows.

Qualifications

Responsibilities

• Build central platform capabilities such as model access, an AI gateway, MCP services and knowledge retrieval, with entitlements, audit and monitoring built in.
• Design, build and run AI assistants, agents and services used by PMs, traders, quants and researchers.
• Stand up and run production inference for open-weight models on GPUs, using engines such as vLLM or SGLang behind OpenAI-compatible APIs.

Job description

Описание

The AI Team, part of Front Office Technology, builds the AI tools and platforms used across the front office: assistants and agents for investment teams, secure access to firm data and knowledge, and the controls that let AI be used safely in a regulated trading environment. We build central capabilities for the whole firm and support teams across the business as they build their own, bringing the most successful into Technology ownership as they grow in importance.

Задачи

Build central platform capabilities such as model access, an AI gateway, MCP services and knowledge retrieval, with entitlements, audit and monitoring built in. Design, build and run AI assistants, agents and services used by PMs, traders, quants and researchers. Stand up and run production inference for open-weight models on our own GPUs, using engines such as vLLM or SGLang behind OpenAI-compatible APIs. Fine-tune models for specific front office tasks with LoRA and QLoRA, and serve multiple adapters efficiently on shared base models. Apply quantisation (e.g. FP8, AWQ, GPTQ) and serving optimisations such as prefix caching, continuous batching, speculative decoding and tensor parallelism to get the most from each GPU. Benchmark latency and throughput (time to first token, inter-token latency, tokens per second) with tools such as GuideLLM, and plan GPU capacity against service levels. Run inference on Kubernetes with GPU scheduling, autoscaling, and monitoring of GPU utilisation and health. Route traffic across hosted and self-hosted models through a central gateway such as LiteLLM, sending each workload to the cheapest model that meets its quality bar. Apply cost levers: model right-sizing, prompt caching, batching and context efficiency. Compare the cost and quality of self-hosted and API models per task, and support provider and hardware decisions with evidence. Identify AI applications built in the business with real usage or value, and bring them onto central platforms. Turn repeated requirements into shared services, MCP servers, skills and agent templates, reducing duplication across teams and providers. Pick up other priorities across the team as they arise, from new front office tools to knowledge services, MCP integrations and data access. Contribute to shared engineering standards, code review and production support. Provide usage, performance and cost metrics from the models and services you run to the colleague responsible for AI spend reporting, so decisions to scale, consolidate or retire AI tools are based on data.

Условия

Build AI tools used in live investment workflows, and help the firm run its own models on its own GPUs. Join an AI function with a real mandate and the backing to shape how AI is built across the firm. Work with current frontier AI tooling in a team that values engineering discipline over hype.