6 months ago
San Francisco, CA, USA or New York, NY, USAMid Level
H1B Sponsor
Base Salary
$180k - $250k/yr
Responsibilities
- Architect and optimize production AI workloads on Modal for customers including Suno, Lovable, Cognition, and Meta.
- Work on LLM serving, model training including SFT and RLHF, audio pipelines, scientific computing, and other demanding workloads.
- Contribute to open-source projects such as SGLang and publish technical content demonstrating Modal’s capabilities.
- Collaborate with product and sales teams as both an engineer and product stakeholder.
- Build trusted relationships with CTOs, engineering VPs, and ML leads.
- Conduct technical demos, experiments, and proofs of concept.
Requirements
- At least 2 years of professional machine-learning engineering experience.
- Hands-on experience in areas such as inference optimization, model training, GPU programming, or ML infrastructure is preferred.
- Familiarity with serving tools such as vLLM and SGLang and training tools such as slime, verl, and TRL; depth in at least one toolchain is sufficient.
- Strong communication skills and the ability to explain technical architecture and tradeoffs to engineering teams and technical leadership.
- Genuine interest in working directly with customers.
- Side projects, open-source contributions, or published work in ML or systems performance are a bonus.
Benefits
- In-person work in New York City, San Francisco, or Stockholm.
- Opportunities to grow within an early-stage, fast-growing AI infrastructure company.
- Opportunity to contribute to open-source projects and publish technical content.
Tech Stack
Seaborn
Categories
Forward DeployedML Engineering
About Modal
Customers rely on Modal for instant GPU access, sub-second container starts, and native storage, so it's simple to serve low-latency inference, fine-tune models, and access production-ready sandboxes at scale. Every era of computing came with new workloads that previous infrastructure couldn't serve: mainframes, databases, the cloud. Each time, the company that rebuilt the layer underneath defined the decade. AI is no different, except it touches everything instead of one slice. The window to build is open right now.