Cross-cloud AI training and inference optimization software

Member of Technical Staff, Post-Training & Applied Research

SF Tensor is building high-performance compute infrastructure, including GPU compilers and model-training systems, to make AI compute faster, cheaper, and more accessible. The Member of Technical Staff will own the post-training modeling lifecycle end to end, including data curation, supervised fine-tuning, reinforcement learning, evaluation, distillation, deployment, and customer collaboration.

Member of Technical Staff, GPU Compiler

SF Tensor is building high-performance compute infrastructure for AI, spanning hardware, compilers, and cloud systems. The Member of Technical Staff will develop and extend a GPU compiler across MLIR, LLVM, backend code generation, search-based optimization, and correctness verification, while improving performance for large-scale training and inference workloads.

Member of Technical Staff, Product Engineering

SF Tensor is building high-performance compute infrastructure and AI systems, including GPU compilers and Model Foundry. The Member of Technical Staff, Product Engineering will own Model Foundry product areas end to end, building researcher-facing interfaces and backend services for experiment orchestration, monitoring, and reproducibility.

Member of Technical Staff, AI-Driven Compilation

SF Tensor is building high-performance compute infrastructure spanning hardware, compilers, and cloud systems for AI workloads. The Member of Technical Staff, AI-Driven Compilation will develop reinforcement learning and agentic compilation systems that search large program spaces, optimize GPU kernels, and integrate learned components into the production compiler pipeline.

Member of Technical Staff, GPU Kernels

SF Tensor is building high-performance compute infrastructure spanning hardware, compilers, and cloud systems to make AI compute faster, cheaper, and more portable. The Member of Technical Staff will write and hand-optimize GPU kernels across NVIDIA, AMD, TPU, and Trainium, profile hardware behavior, and convert kernel optimization insights into compiler-searchable structures and performance models.

Member of Technical Staff, Sandbox Infrastructure

SF Tensor is building high-performance compute infrastructure for AI, including GPU compilers and systems for running workloads across different vendors and cloud environments. The Member of Technical Staff, Sandbox Infrastructure will build and operate a serverless GPU container service supporting secure, high-fidelity execution across accelerators, vendors, spot capacity, multiple nodes, and live migration.