AI Research Engineer, Inference
An engineering role on HRT's AI Labs (HAIL) team, which builds the large models many trading teams rely on. The role covers making large-model inference fast across GPUs, FPGAs and ASICs.
What you’d work on
Our reading of the posting: what a new hire in this role would most likely spend their first months on.
- Write and tune GPU kernels for model inference
- Explore specialised inference hardware such as FPGAs and ASICs
- Improve data streaming into live models
- Co-design model changes with researchers to meet latency goals
What they ask for
Required
- GPU kernel skills (CUDA, Triton, Pallas or CuTe)
- Low-level PyTorch, JAX or XLA work
- Two or more years building deep learning systems in any field
- Ability to move methods between domains
Preferred
- LLM experience
- FPGA or ASIC experience
- Education
- No degree level stated.
- Experience
- 2+ years building deep learning systems; finance experience not required
- Languages and tools
How to prepare
- Write and benchmark a CUDA or Triton kernel for attention or matrix multiply
- Study inference optimisations such as batching, quantisation and CUDA Graphs
- Learn the basics of FPGA-based inference
- Profile a deployed model and cut its latency
What to study
- Python and Data Primer
- Financial Machine Learning
- Trading Systems and Low-Latency Infrastructure
- Modern C++ for Quantitative Finance
The HFT Quant Developer roadmap sets out the stages for this kind of role, with free resources for each.
Open the roadmapApply
Read the full posting on Hudson River Trading’s site before applying: requirements, location and pay can change after we check them.
View the posting on Hudson River Trading’s site ↗
SelfMadeQuant wrote this summary from the employer’s public posting, last checked 6 November 2026. SelfMadeQuant isn’t affiliated with Hudson River Trading and doesn’t handle applications.