Forward Deployed Engineer - ML
New York • FullTime
Posted 6mo ago
About the job
We are seeking Forward Deployed ML Engineers to work at the intersection of deep technical work and direct customer impact. As an ML FDE, you will partner with leading AI companies and foundation model labs to help them achieve state-of-the-art performance on their most demanding workloads, including LLM serving, model training (SFT, RLHF), audio pipelines, and scientific computing. You will help teams reach outcomes that many engineers cannot achieve on their own. The FDE team comprises world-class software engineers, computational scientists, ML engineers, and former founders, and we are looking for individuals with strong engineering fundamentals, deep curiosity across the AI stack, and energy for tackling hard problems directly with customers.
Responsibilities
- Work hands-on with companies to architect and optimize production AI workloads on Modal.
- Contribute to open-source projects and publish technical content demonstrating Modal's capabilities.
- Collaborate with Modal's product and sales teams, acting as both an engineer and a product stakeholder.
- Build trusted relationships with technical leaders at companies engaged in frontier AI work.
- Conduct technical demos, experiments, and proof-of-concepts to showcase Modal's performance advantages.
Requirements
- 2+ years of professional ML engineering experience, with hands-on work in inference optimization, model training, GPU programming, or ML infrastructure.
- Familiarity with serving (e.g., vLLM, SGLang) and training (e.g., slime, verl, TRL) toolchains, with the ability to go deep on at least one.
- Strong communication skills, capable of discussing technical architecture with engineering teams and articulating tradeoffs to leadership.
- Genuine interest in working directly with customers and solving their problems.
- Willingness to work in-person in New York City, San Francisco, or Stockholm.
- Bonus: side projects, open-source contributions, or published work in ML or systems performance.