Developer Advocate, MAX Inference & Serving

• Join our world-leading product team and be part of redefining how AI infrastructure is built and deployed.
• Additionally, this role requires travel to conferences and developer events, which may be as often as once per month, as well as travel for team and company events (typically 2-4 times per year).
• Build and publish reproducible benchmarks comparing MAX against vLLM, SGLang, TensorRT-LLM, and Triton Inference Server, including methodology, harness code, and hardware configurations so others can verify the numbers.
• Deploy and profile real inference workloads on MAX across CPU and GPU targets, and investigate performance gaps in latency, throughput, and cost per token.
• Write and maintain technical content grounded in that work: performance deep dives, serving architecture explainers, and posts that show how MAX handles batching, KV cache management, quantization, and multi-GPU serving.
• Author runnable tutorials, examples, and video walkthroughs covering model deployment on MAX, from pip install modular to a served endpoint under load.
• Provide technical support to engineers evaluating MAX for inference and serving, including teams at some of the world's largest companies, and debug their deployment and performance issues directly.
• Answer technical questions across GitHub, Discord, X, and LinkedIn, reproducing reported issues and filing them with enough detail for engineering to act on.
• Translate community and customer findings into prioritized inference and serving feedback for engineering and product, shaping the MAX roadmap.
• Present technical talks, benchmark results, and live demos at conferences, summits, and meetups.
• Help set the technical direction of Modular's developer relations content, including what gets benchmarked, documented, and demoed next.
• 3-5 years of professional engineering experience, with at least some of it spent deploying or operating ML systems in production. You have run inference workloads yourself, not just written about them.
• Strong Python and systems programming experience in C++, Rust, Mojo, or CUDA is a significant advantage, especially if you have profiled and optimized GPU code.
• Comfort with the deployment surface around serving: containers, Kubernetes, GPU drivers and runtimes, and the usual ways a cluster refuses to cooperate.
• You benchmark rigorously. You know why one number is not a result, how to control for warmup and batch size, and when a comparison is not apples to apples.
• You write well about technical work, and you have a portfolio to show it: blog posts, tutorials, videos, docs, or courses. You explain complex ideas without losing precision, and you cut the filler.
• You learn new tools fast and turn them into accurate content quickly. A feature ships Tuesday, your tutorial goes out Thursday and the code runs.
• A growth and leadership mindset, with a collaborative attitude that seeks to learn more from our customers, team members, and the broader market
• Bachelor's degree in Engineering, Information Systems, Computer Science, or technical related field.
• 2+ years of Product Management or related work experience.
• Completed advanced degrees in a relevant field may be substituted for up to two years of work experience.
• Experience programming GPUs using CUDA or ROCm.
• Familiarity with the Mojo 🔥 programming language and MAX AI framework.
• Familiarity with open source software development practices and communities.

None flagged.