Research Engineer, Benchmarking - Member of Technical Staff

Callosum Technologies Ltd.; Greater London; Hybrid; GBP 75,000 - 120,000; Full time

Benefits offered by this job

Visa sponsorship

Relocation benefits

Job summary

Callosum Technologies Ltd. is hiring for a research-driven role in London to build and run a unified benchmarking system for evaluating agentic and algorithmic LLM workflows.

You will design tasks, collect real commits/traces, and ensure robust, contamination-free evaluation with reproducible results. The role requires a PhD or equivalent track record, strong Python, and experience with sandboxed/distributed execution.

Qualifications

• PhD in computer science, ML, or related field, or equivalent research track record.
• Authorship of benchmark/evaluation papers at NeurIPS/ICML/ACL is preferred.
• Understanding of evaluation pitfalls: contamination, overfitting, irreproducible results.
• Strong Python programming and experience with sandboxed/distributed execution and CI.
• Hands-on experience building or evaluating multi-step LLM systems.

Responsibilities

• Design and build a unified benchmarking system for evaluating agentic and algorithmic workloads.
• Mine real commits/traces, run sandboxed grading, and build task suites reflecting real agentic work.
• Enforce controls against contamination, overfitting to benchmarks, and metric gaming.
• Compare algorithmic and agentic approaches honestly, focusing on resolved-task quality.
• Lead external benchmark co-publications and publish results.
• Incorporate results into which approach ships and QA quality claims.

Skills

Benchmark evaluation

Sandboxed execution

Python

Agentic/LLM eval

Publications in benchmarks

Education

PhD in computer science or related field

Tools

Python

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

Choosing between algorithmic strategies for multi‑step LLM work is a measurement problem, and most teams solve it badly: comparisons run case by case, by whoever needs them that week, on whatever task is closest to hand. That doesn’t scale, and it doesn’t hold up to outside scrutiny – from a customer, or from a reviewer. Callosum needs one benchmarking system: reproducible, contamination‑controlled, and trusted enough to be the evidence that decides which approach ships.

This role owns that system. You will build a harness that measures task success, quality, and robustness across motifs, agent topologies, and decomposition strategies, grounded in execution – real commits, real traces, sandboxed grading – rather than self‑reported or model‑graded scores. The results become the proof points we show customers, the evidence behind the benchmarks we co‑publish, and the basis on which an approach ships or doesn’t.

This is a research hire that builds. We expect the rigour of a strong evaluation paper applied to a production system, and the engineering ability to design, build, and curate it yourself rather than hand it off.

What You’ll Build

• Design and build a unified system for evaluating agentic and algorithmic solutions – task success, quality, and robustness across motifs, agent topologies, and decomposition strategies, on workloads that match what customers actually run. Cost per resolved task is an outcome you track, not the object of the exercise.

• Mine real commits and traces, run sandboxed execution grading, and build task suites that reflect real agentic work: code search, code edit and repair, repository summarisation, tool use. Self‑reported or model‑graded success isn’t enough on its own.

• Enforce controls against contamination, over‑fitting to benchmarks, and metric gaming, and keep baselines stable over time – any result should be re‑runnable to the same number, by us or by a reviewer

• Compare algorithmic and agentic approaches honestly, not models or chips – a motif that adds steps, latency, or cost has to earn it in resolved‑task quality, and the system says clearly when it doesn’t

• Lead external benchmark co‑publications, held to a standard that survives peer and customer review

• Feed results directly into which approach ships, into the proof points behind customer engagements, and review quality claims across the company before they go out

What You’ll Bring

• PhD in computer science, machine learning, or a related field, or an equivalent research track record

• Authorship or co‑authorship of a benchmark or evaluation paper at a recognised venue – NeurIPS Datasets and Benchmarks, ICML, ICLR, ACL – ideally on agentic or LLM evaluation, or a comparably rigorous evaluation contribution

• A working understanding of how LLM and agent evaluation goes wrong: contamination, over‑fitting to benchmarks, weak baselines, under‑powered comparisons, irreproducible results

• The engineering ability to design, build, and curate these systems decisively – strong Python, and comfort with sandboxed and distributed execution and CI

• Hands‑on experience building or rigorously evaluating agentic or multi‑step LLM systems

What Sets You Apart

• Published agentic or tool‑use benchmarks that use execution‑based grading

• Experience running sandboxed execution grading at scale

• Open‑source evaluation or harness tooling

• Familiarity with code‑agent workloads such as search, edit, and repair

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.