Inference System & Performance - Member of Technical Staff

Callosum; Greater London; On-site; GBP 101,000 - 192,000; Full time

Benefits offered by this job

Competitive salary

Equity & Ownership

Private healthcare

Visa sponsorship & relocation

London office on-site

Job summary

Callosum in London is hiring for a role owning end-to-end performance for our inference platforms. You will manage KV caches, batching, memory, and scheduling across heterogeneous hardware to scale model serving.

The candidate should have deep LLM inference knowledge, strong distributed system experience, and low-level debugging skills on GPUs, networks, and Linux. Visa sponsorship and relocation are available; on-site in London.

Qualifications

• Deep understanding of LLM inference internals: KV cache lifecycle, attention mechanisms, and serving architectures.
• Strong systems engineering background with proven experience optimising distributed GPU workloads.
• Proficiency in C++, CUDA, Python, Rust, or similar - and the instinct to go low-level when it matters.
• Hands‑on debugging skills across GPU, networking, and Linux systems - able to work from first principles with limited tooling.

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 internals

Distributed GPU workloads

Cpp/CUDA/Python

Linux performance analysis

Tools

NCCL

NVLink

RDMA

InfiniBand

RoCE

Job description

About Us

We’re living through a Cambrian explosion of intelligence: new models and new chips, each specialised for different tasks, are arriving all at once. The result is a new era for AI, one of radical heterogeneity.

Callosum is the Intelligent Systems Company. We believe the next generation of AI won't be defined by any single model or chip, but by intelligent systems in which hardware and intelligence co-evolve. We are building the infrastructure that unifies heterogeneous compute across the full stack. This opens a new axis of scaling intelligence: a dynamic system that tailors itself to what each workload actually needs, whether that's speed, cost, precision, or whatever unit comes next.

The last era scaled on a different bet: one bigger model, more of the same chip, more data. That bet is running into structural limits. Frontier models offer extraordinary capability at unsustainable cost, one that today's monolithic infrastructure was never designed to serve.

Our founding principle is that intelligence comes from many specialised systems working together, not from any single component. We build the software orchestration layer that co-evolves models, workflows and silicon into one system, delivering inference tailored to every workload, and demonstrating orders‑of‑magnitude leaps in capability and cost.

Because our software spans the full stack, our engineering team works directly with heterogeneous accelerators and frontier silicon, including Cerebras, d‑Matrix, Intel, NVIDIA, AMD, Normal Computing, Tenstorrent, GreatSky, and Mixx. We are not stopping at today's chips: each new generation of silicon unlocks algorithms that couldn't run before, and we intend to be first to them, every time. If we get it right, it will belong to everyone building on it - not to any single vendor.

In our latest funding round, we raised $100M, led by Atomico with participation from Plural, DCVC and the UK Sovereign AI Fund’s first investment. With this, we are building the infrastructure for the next era of intelligence.

We are engineers and scientists based in London, working across the full depth of the stack. We are curious, intellectually honest, and building what doesn't exist yet. If you thrive on uncharted territory and are energised by the scale of the challenge, we'd love to hear from you.

About the Role

Standard inference architectures typically focus on monolithic chip types and model classes. Callosum intentionally breaks this mold, operating heterogeneous hardware at scale across a diverse model portfolio. Success in this environment requires an inference layer built entirely from first principles.

Sitting at the heart of our technical mission, this position owns end-to-end performance for our inference platforms. Your focus will span KV cache strategies, batching internals, memory management, and multi-node scheduling. You will develop the core software driving execution speed, silicon efficiency, and platform scalability as we expand our hardware and model footprint. This is a high-leverage role tackling complex system challenges across the entire stack.

What You’ll Build

• 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

What you Bring

• Deep understanding of LLM inference internals: KV cache lifecycle, memory management, attention mechanisms, and serving architectures

• Strong systems engineering background with proven experience optimising distributed GPU workloads

• Proficiency in C++, CUDA, Python, Rust, or similar - and the instinct to go low-level when it matters

• Hands‑on debugging skills across GPU, networking, and Linux systems - able to work from first principles with limited tooling

What Sets You Apart

• Experience building or significantly optimising production‑grade, high‑throughput model serving stacks

• Multi‑GPU and multi‑node inference optimisation using NCCL, NVLink, RDMA, InfiniBand, or RoCE

• GPU memory profiling, CUDA or Triton kernel optimisation

• Linux performance analysis and optimisation

What We Offer

• Competitive Salary, determined by skills and experience

• Equity & Ownership

• Private healthcare

• We offer Visa sponsorship and relocation benefits to hire the best in the world

• We work in person at our London office. You'll have the tools, space and setup to do your best work, and if you have specific needs, just tell us

We're committed to building an inclusive workplace where everyone feels welcome, and believe in equal opportunities for all.

Compensation: £101K – £192K Offers Equity