High Performance Engineer (up to £140k)

Dex; Cambridge; On-site; GBP 90,000 - 130,000; Full time

Job summary

Dex in Cambridge seeks a senior IC to own performance across GPU physics simulation, policy learning loops, and edge execution for humanoid and quadruped robots. You profile systems, reclaim milliseconds, and collaborate daily with founders who still write engine code.

You will optimize memory layouts, parallel execution paths, and test gains on lab robots, while setting benchmarks and driving architectural decisions for the core engine.

Qualifications

• Mastery of modern C++ with a track record of faster systems.
• Experience in latency- or throughput-driven environments like game/physics engines, HPC, or real-time systems.
• Ability to raise the technical bar through rigorous code reviews and pairing.

Responsibilities

• Profile and optimize GPU and CPU execution pipelines to eliminate latency across the simulation engine, training loops, and edge deployment.
• Architect cache-friendly memory layouts, lock-free structures, and parallel execution paths tailored to modern hardware limits.
• Test and validate performance gains on physical quadruped and humanoid robots in the lab.
• Set the engineering standard for performance testing and benchmarking through hands-on pairing and code review.
• Work with founders to debate architecture and make irreversible design calls on the core engine.

Skills

Modern C++

Performance profiling

Parallelism

Hardware awareness

Job description

The role

Most robotics simulation hits a bottleneck because the physics engine cannot run fast enough to train policies at scale. This team builds GPU-accelerated simulation and training infrastructure where speed is the entire value proposition, founded by two engineers who spent over a decade building physics engines at the world's dominant GPU architecture company. Backed by top-tier VCs with extensive runway, they run a lean 11-person team in Cambridge with physical hardware in the room.

You will own performance across the entire stack: from GPU physics simulation and policy learning loops to real-time edge execution on humanoid and quadruped robots. This is a senior IC seat, not a middle-management layer. You are here to profile systems, reclaim milliseconds alongside founders who still write engine code daily, and provide the technical anchor that a sharp, early-career team needs in the room.

The work

• Profile and optimize GPU and CPU execution pipelines across the simulation engine, training loops, and edge deployment to systematically eliminate latency.
• Architect cache-friendly memory layouts, lock-free structures, and parallel execution paths tailored to modern hardware limits.
• Test and validate performance gains directly on physical quadruped and humanoid robots inside the lab.
• Set the engineering standard for performance testing and benchmarking across the team through hands-on pairing and code review.
• Work directly with the technical founders to debate architecture and make irreversible design calls on the core engine.

What You Bring

• Mastery of modern C++ with a demonstrable record of making systems measurably faster through low-level profiling, memory layout design, parallelism, and deep hardware awareness.
• Prior work in environments where latency or throughput is the product itself, such as game or physics engines, HPC, scientific simulation, low-latency trading, or embedded real-time systems.
• Proven ability as a senior individual contributor to raise the technical bar of an early-career engineering team through rigorous code reviews, direct pairing, and clear task decomposition.
• Ability to work fully on site at the lab in Cambridge, UK every day.