ML Infrastructure Engineer

Job description

**About zaimler** AI agents can't reason over data they don't understand. Enterprise data today is fragmented across dozens of systems with no shared context, meaning, or structure, and that's why most enterprise AI is failing. The shift from copilots to autonomous agents is creating an entirely new infrastructure layer, and we're building it. zaimler is the context infrastructure for the agentic era: a platform that automatically discovers domain knowledge, maps relationships, and gives AI agents the semantic understanding to operate with precision at scale. Imagine knowledge graphs that support real-time inference, built for systems that need to reason, not just retrieve. zaimler was founded by Biswajit Das (ex-VP Engineering, Truera), a Data Infra veteran and former Chief Architect at Visa, and Sofus Macskassy (ex-Director of Engineering, LinkedIn), who built one of the largest knowledge graphs in production in the industry at LinkedIn. We're growing and deploying with major enterprises across insurance, travel, and technology. If you want to build infrastructure that the next decade of enterprise AI runs on, we'd love to talk. ### **About the Role** You’ll join our ML team focused on turning raw enterprise data into structured, contextualized knowledge graphs and embeddings. You’ll develop novel and highly scalable algorithms for ML and data engineering to make our overall system more efficient, experiment with new approaches for distilling large models into smaller, more efficient ones; improve retrieval, ranking, and reasoning performance through feedback loops; and prototype methods that help LLMs extract and act on real-world knowledge. We're looking for someone who thrives on iteration, cares about building with rigor, and is hungry to learn from some of the best engineers and researchers in the field. ### What You’ll Be Doing - Build and maintain training infrastructure, feature stores, and model serving pipelines - Optimize LLM inference performance — compute efficiency, memory management, latency, and throughput - Read, debug, and contribute to LLM runtime and supporting library code (Rust and/or C++) - Deploy and manage models at scale using tools like vLLM and Baseten - Architect scalable pipelines for model training and serving across GPU infrastructure - Collaborate with ML and data engineers to ensure the platform meets research and production needs ### Prior Experience - PhD in CS, ML, or a related field or MS with 4+ years of relevant industry experience - Background in LLM optimization: inference efficiency, quantization, memory layout, or serving performance - Ability to read, navigate, and debug LLM source code and underlying runtime libraries - Comfortable in Rust and/or C++ at the systems level; strong Python required - Strong algorithmic fundamentals — data structures, complexity, distributed systems - Hands-on experience with model serving infrastructure (vLLM, Baseten, Triton, or similar) - Experience setting up and scaling ML pipelines end-to-end ### Nice to Have - Familiarity with feature store design and management - Experience with GPU cluster management and optimization - Contributions to open-source ML infrastructure or LLM tooling - Experience with Ray, ONNX, TensorRT, or similar optimization and serving frameworks - Understanding of transformer internals and attention mechanisms at the implementation level ### Why Join - A rare chance to be a founding engineer shaping both company and product direction. - Competitive salary, benefits, and meaningful equity. - Work alongside engineers and researchers from LinkedIn, Visa, Meta, and Branch. - Onsite culture in San Mateo, designed for deep collaboration and high-velocity building. - Full benefits package (Medical, Dental, Vision, 401k). - We sponsor H-1B visas and assist with immigration processes.