ML Research Scientist - Inference ML
cerebras · United States and Canada
Description du poste
About the role
Cerebras Systems builds the world's largest AI chip, delivering training and inference speeds over 10 times faster than GPU-based cloud services. As an ML Research Scientist on the Inference ML team, you will invent and evaluate algorithms, methods and model architectures for efficient inference on the Cerebras architecture, working with researchers and engineers to design, prototype, validate and optimize models. The role focuses on speculative decoding, large-model pruning and compression, sparse attention and sparsity-driven techniques for low-latency, high-throughput inference. It is a hybrid position based in Toronto, Ontario, Canada or Sunnyvale, California, USA.
Key responsibilities
- Research and prototype inference algorithms and model architectures that exploit Cerebras hardware, focusing on speculative decoding, pruning/compression, sparse attention and sparsity.
- Design and evaluate new transformer architectures and variants for NLP and computer vision to balance quality, latency, throughput and compute.
- Pursue high-impact research at the intersection of model architecture, inference systems and hardware.
- Train or adapt models and measure inference performance with Cerebras tools to maximize throughput and minimize latency.
- Collaborate with engineering and performance teams to turn research into usable systems, from inception through delivery.
- Publish and present research findings and contribute to open-source releases where appropriate.
Required profile
- PhD in machine learning, computer science or a closely related field with 3+ years of relevant research or industry experience, or a master's degree with an exceptional record of original research and 4+ years of experience.
- Deep understanding of modern ML architectures, generative models in language and/or vision, and inference behavior.
- Evidence of research impact through publications in leading conferences or journals, high-quality preprints, or widely used open-source contributions.
- Preferred: original research in speculative decoding, pruning, compression, sparse attention, quantization, MoE models, sparsity or inference-aware design; experience owning complex ML projects end-to-end; experience designing novel model architectures; familiarity with large-scale model training and deployment.
Required skills
- Strong Python programming and ML frameworks such as PyTorch, Transformers, vLLM or SGLang.
- Ability to reason about model and algorithm performance for inference workloads.
- Preferred: experience optimizing performance on specialized hardware (GPUs, TPUs, HPC interconnects); Triton, CUDA or C++ experience is a plus.
Questions fréquentes
Pourquoi signalez-vous cette offre ?
Aller plus loin
Salaires, guides et recherches au Canada.
Salaires par métier
Postulez en 30 secondes
Entrez votre email pour postuler. Un compte sera cree automatiquement.
En continuant, vous acceptez nos conditions d'utilisation.
Deja un compte ? Connexion
Une question sur cette offre ?
Posez-la ici : vous recevrez le récapitulatif de l'offre par e-mail, tout de suite.
Publie il y a 7 heures
Expire dans 1 mois
3 vues · 0 interesses
Boostez vos chances
Importez votre CV : nous vous proposons les offres qui matchent votre profil.
Analyse de votre CV en cours...
cerebras
United States and Canada
Offres similaires
-
Principal AI Security Engineer
cerebras United States and Canada -
Software Engineer, Kernel Reliability
cerebras United States and Canada -
Technical Lead Manager, Infrastructure Hardware (Server and Network Systems)
cerebras United States and Canada -
Principal Engineer, Identity & Access Management
Questrade Financial Group -
Technical Support Specialist - Telecom
servicetitan Canada Manitoba Remote