← All roles

ML Systems & Performance Engineer

Engram· Member of Technical Staff
Apply on company site
San FranciscoFull-timePosted 4mo ago

About the role

Role focused on designing, optimizing, and scaling training and inference workloads to bridge cutting-edge AI research and production, with a strong emphasis on personalization, continual learning APIs, and collaboration between researchers, platform engineers, and customers.

Responsibilities

  • Designing and executing new frameworks, techniques, and systems to improve performance, reliability, latency, and efficiency.
  • Partner closely with researchers, turning prototypes into systems that run at scale and feeding systems constraints back into research decisions.
  • Optimize serving paths for personalization and memory retrieval, where per-user state and low latency both matter.
  • Work on distributed training — data and model parallelism, communication scheduling, and scaling efficiency across multiple GPUs and nodes.

Qualifications

  • Bachelor’s degree or equivalent experience in computer science, engineering, or similar.
  • 5+ years of experience with training or inference systems, optimized workloads with measurable results.
  • Strong engineering foundation, with demonstrated excellence navigating complex technical environments and shipping high-quality code in a fast-paced environment.
  • Deep understanding of ML framework (e.g. PyTorch, JAX), GPUs, distributed systems, and infrastructure.
  • Operate well in ambiguous environments — you will have real ownership and be responsible for steering the ship in a novel sector of the industry.
  • You have a bias toward action and a knack for turning research concepts into concrete, executable plans.

Skills

Required:Bachelor’s degree or equivalent experienceExperience with training or inference systems

Benefits

  • Competitive cash compensation and startup equity.
Last updated Aug 4, 2026. Interested? Apply on the company site.

About Engram

AI has learned the world. Now it should learn yours.

Today’s AI models don’t understand what you do. Not really. Everything models know comes from their training – and they’re trained mostly on the public internet.

NotionHarveyMicrosoftengram modelsM365Custom AgentsNotion workspacesAI context
Company profileWebsite

More open roles at Engram