Cerebras
ML Research Scientist - Inference ML
United States and Canada · Full-time · Posted Oct 2, 2026
- Experience
- PhD route: 3+ years of relevant research or industry experience; master’s route: 4+ years of relevant experience.
- Education
- PhD in machine learning, computer science, or a closely related technical field, or a master’s degree with an exceptional record of original, independent research.
About the job
Develop and evaluate algorithms, model architectures, and methods that improve inference efficiency on Cerebras hardware.
What you'll do
- Research and prototype inference algorithms and transformer architectures, including speculative decoding, pruning, compression, sparse attention, and sparsity techniques.
- Train or adapt models and evaluate inference performance to improve quality, latency, throughput, and compute trade-offs.
- Collaborate with engineering teams to translate research into usable systems and publish or present findings.
What they ask for
- PhD in machine learning, computer science, or a closely related technical field, or a master’s degree with an exceptional record of original, independent research.
- Deep understanding of modern ML architectures, generative models, ML systems, and inference behavior.
- Evidence of relevant research impact through publications, preprints, or widely used open-source contributions.
- Strong Python skills and proficiency with relevant ML frameworks.
Nice to have
- Research in speculative decoding, pruning and compression, sparse attention, quantization, mixture-of-experts, sparsity, post-training, or inference-aware design.
- Experience with Triton, CUDA, or C++.
Tools
Python, PyTorch, Transformers, vLLM, SGLang, Triton, CUDA, C++
AI focus
Research and optimize AI-model inference, including speculative decoding, model pruning and compression, sparse attention, and inference-efficient architectures.
Model-written summary · the original posting is authoritative