Senior Inference Systems Engineer – Multi-Node GPU

Callosum; Greater London; On-site; GBP 120,000 - 160,000; Full time

Benefits offered by this job

Equity

Private healthcare

Visa sponsorship

Relocation benefits

London office

Job summary

Callosum, based in London, is seeking an experienced infrastructure-focused engineer to own end-to-end performance for its heterogeneous inference platforms. The role spans KV cache strategies, batching, memory management and multi-node scheduling to drive speed and efficiency across a growing model portfolio.

You will design, profile, and optimise GPU kernels and networking paths, build tooling for visibility, and collaborate across the stack to scale hardware and models.

Qualifications

• Experience with KV cache lifecycle, memory management, attention mechanisms, and serving architectures.
• Proven track record optimising distributed GPU workloads.
• Proficiency in C++, CUDA, Python, Rust, or similar.
• Hands-on debugging across GPU, networking, and Linux systems.

Responsibilities

• Design and optimise inference serving systems across heterogeneous multi-GPU and multi-node environments.
• Own KV cache lifecycle management, batching strategies, and memory allocation to maximise throughput and minimise latency.
• Profile and tune GPU kernels, identify bottlenecks across compute, memory, and network, and implement targeted optimisations.
• Build and improve scheduling logic for continuous batching, disaggregated prefill/decode, and speculative decoding.
• Work with networking primitives - NCCL, NVLink, RDMA, InfiniBand, RoCE - to optimise communication across distributed inference workloads.
• Develop tooling for performance visibility, regression detection, and benchmarking across hardware configurations.

Skills

LLM inference

Systems engineering

C++

CUDA

Python

Rust

Linux

Tools

NCCL

NVLink

RDMA

InfiniBand

RoCE

Job description

Callosum, based in London, is seeking an experienced infrastructure-focused engineer to own end-to-end performance for its heterogeneous inference platforms. The role spans KV cache strategies, batching, memory management and multi-node scheduling to drive speed and efficiency across a growing model portfolio.

You will design, profile, and optimise GPU kernels and networking paths, build tooling for visibility, and collaborate across the stack to scale hardware and models.