Ml Model Serving Engineer • Senior

Who you are

You are an expert in differentiable array computing frameworks, preferably PyTorch, with a strong background in optimizing machine learning models for serving reliably at high throughput and low latency. Your significant systems programming experience means you are comfortable with the internals of high-performance server systems, and you have a knack for modifying and extending LLM serving frameworks like VLLM and SGLang to leverage the latest techniques in model serving. You understand the importance of reducing model initialization times without sacrificing quality, and you are familiar with techniques like in-flight batching, caching, and custom kernels to speed up inference.

What you'll do

In this role, you will turbocharge Sesame's serving layer, which consists of a variety of LLM, speech, and vision models. You will partner closely with ML infrastructure and training engineers to build a fast, cost-effective, accurate, and reliable serving layer that powers a new consumer product category. Your responsibilities will include identifying opportunities to produce faster models without sacrificing quality and implementing optimizations that enhance the overall performance of the serving layer. You will also collaborate with the training team to ensure that the models are not only efficient but also maintain high standards of accuracy and reliability.

What we offer

Sesame provides a full-time position with competitive benefits, including 401k matching, 100% employer-paid health, vision, and dental benefits, unlimited PTO, and flexible spending account matching. You will be part of a team that is shaping the future of lifelike computers, working alongside founders from Oculus and Ubiquity6, and leaders from Meta, Google, and Apple. We encourage you to apply even if your experience doesn't match every requirement, as we value diverse perspectives and backgrounds.