ML Research Scientist in Inference at Cerebras Systems (Toronto or Sunnyvale)

Job Description

This ML research scientist job in inference is with Cerebras Systems, the maker of a wafer-scale AI chip. You join the Inference Core team. You will look for new algorithms and model designs that make inference fast on Cerebras hardware.

At a glance

• Position: ML Research Scientist, Inference ML, full time
• Where: Hybrid, in Toronto (Canada) or Sunnyvale, California (USA)
• Team: Inference Core
• Pay: Not stated in the posting
• Posted: 2 October 2026
• Apply: open until filled (re-checked 14 November 2026)

About the inference ML research

• Research and prototype new inference algorithms and model designs that use Cerebras hardware well
• Work on speculative decoding, pruning and compression, sparse attention and sparsity
• Design and test new transformer variants for language and vision, balancing quality, speed and compute
• Train or adapt models and measure their speed with Cerebras tools
• Work with engineering and performance teams to turn research into working systems
• Publish papers and contribute to open source

Why this role matters

Inference now costs more than training for many AI products. A chip with a different design calls for different algorithms. Here you can shape model ideas around hardware that few researchers can access.

What Cerebras is looking for

• A PhD in machine learning, computer science or a close field plus 3 years of research or industry experience; or a master’s with an outstanding record of original research plus 4 years
• Deep knowledge of modern ML architectures, generative models and ML systems
• Proof of research impact: papers, preprints or widely used open-source work
• Strong Python and PyTorch, and experience with Transformers, vLLM or SGLang

Nice to have

• Original research in speculative decoding, pruning, sparse attention, quantisation, mixture-of-experts or post-training
• End-to-end ownership of ML projects and design of new architectures
• Performance work on GPUs, TPUs or HPC systems, and large-scale training
• Triton, CUDA or C++

Pay and location

The posting gives no salary figure and does not mention visa support. The role is hybrid, based in Toronto or Sunnyvale.

How to apply

Apply through the official Cerebras job posting. Cerebras’s posting does not give a closing date, so the role is open until filled; ResearchJobs.in will re-check this listing on 14 November 2026. Check the posting for location and work-authorisation details before you apply.

See all our artificial intelligence jobs, or browse more research jobs on ResearchJobs.in.