Machine Learning Performance Engineer

JR United Kingdom; City Of London; On-site; GBP 60,000 - 100,000; Full time

Job summary

Une nouvelle startup à Londres recherche un Ingénieur en Performance d'Apprentissage Automatique pour rejoindre leur équipe technique. Le candidat idéal aura une expertise en ingénierie GPU et un esprit de collaboration pour résoudre les défis en intelligence artificielle. Une opportunité unique de façonner l'avenir d'une entreprise innovante.

Qualifications

• Compétences en ingénierie solides en Python, C++, ou Rust.
• Expérience prouvée avec l'ingénierie de performance GPU.
• Familiarité avec des frameworks ML comme PyTorch.

Skills

Python

C++

Rust

CUDA

Tensor Cores

Profiling and debugging

Distributed systems

Job description

Social network you want to login/join with:

Machine Learning Performance Engineer, london (city of london)

col-narrow-left

Client:

Location:

london (city of london), United Kingdom

Job Category:

Other

-

EU work permit required:

Yes

col-narrow-right

Job Views:

3

Posted:

16.06.2025

Expiry Date:

31.07.2025

col-wide

Job Description:

We are recruiting on behalf of an ambitious new startup founded by an exceptional team of ex-big tech researchers and engineers. Based between London and SF, they’ve recently raised over $15M in pre-seed funding from world-class investors and are building a technical founding team to take on some of the hardest and most exciting challenges in AI today.

The company is still in stealth, but their focus is bold and clear: pushing the boundaries of foundational model architecture, efficient training at scale, and real-world deployment of intelligent agents. This is a rare opportunity to join early and shape the technical DNA of a company that is making major mark in the future of AI/AGI.

What We’re Looking For

There’s no checklist, but you’ll likely thrive in this role if you have:

Technical Experience

• Strong engineering skills in Python, C++, or Rust
• Proven experience with GPU performance engineering: CUDA, PTX/SASS, Tensor Cores, memory hierarchy, warp-level primitives
• Familiarity with ML frameworks like PyTorch, and their internals
• Proficiency in profiling and debugging tools like NSight, CUDA GDB, nvprof, NSight Compute
• Deep knowledge of Triton, cuDNN, cuBLAS, CUTLASS, CUB, or similar libraries
• Experience optimising across the stack: from kernel-level compute to cluster-wide networking and memory IO

Systems Fluency

• Background in distributed systems or HPC: understanding of Infiniband, NVLink, RoCE, GPUDirect, NCCL, MPI
• Experience with multi-node training, collective communication algorithms, and throughput analysis
• Comfort navigating complex systems to answer questions like: “Is this a memory bandwidth ceiling or a kernel launch inefficiency?”

Your Mindset

• A hacker’s curiosity: you love breaking things down and figuring out how to make them faster
• Product intuition: performance isn’t abstract to you, it’s about real-world impact
• Collaborative spirit: you’re excited to work across research, infra, and open-source teams
• A bias toward open science, transparency, and high-integrity work

At Adamas Knight, we are committed to creating an inclusive culture. We do not discriminate based on race, religion, gender, national origin, sexual orientation, age, veteran status, disability, or any other legally protected status. Diversity is highly valued, and we encourage applicants from all backgrounds to apply.