About the job
Our team focuses on enabling custom models and dedicated inference on Together. We are responsible for building a container platform, optimizing autoscaling, minimizing cold starts, achieving the best end-to-end model performance, and providing a best-in-class developer experience with great tooling. We often focus on video or audio generation across the stack: CUDA kernels, pytorch optimization, inference engines, container orchestration, queueing theory, etc. An ideal candidate will be great at profiling/optimization but know the word kubernetes, or be intimately familiar with multi-cluster scheduling and have some sense of ML bottlenecks.
Responsibilities
New hires may work on multi-cluster orchestration, portfolio optimization, predictive autoscaling, control panes, model bring-up, model optimization, APIs for managing deployments, inference worker SDKs, and CLI tools.
Analyze and improve the robustness and scalability of existing distributed systems, APIs, databases, and infrastructure
Partner with product teams to understand functional requirements and deliver solutions that meet business needs
Write clear, well-tested, and maintainable software and IaC for both new and existing systems
Conduct design and code reviews, create developer documentation, and develop testing strategies for robustness and fault tolerance
Qualifications
Minimum
5+ years of demonstrated experience in building large scale, fault tolerant, distributed systems.
Excellent understanding of low level operating systems concepts including concurrency, networking and storage, performance and scale
Expert-level programmer in one or more of Python, Golang, Rust, C++, or Haskell
Proficiency in writing and maintaining Infrastructure as Code (IaC) using tools like Terraform
Experience with Kubernetes internals or other container orchestration systems
Bachelor’s or Master’s degree in Computer Science, Computer Engineering, or a related technical field, or equivalent practical experience
Preferred
Experience running serverless inference platforms, doing model bring-up on short notice, being on call, or running a cloud provider is a very big plus
Good taste and ability to thoughtfully discuss how what you’ve built has failed over time
Experience designing, analyzing and improving efficiency, scalability, and stability of various system resources
Sound judgement for when to use and when to not use LLMs for code
Writing-heavy roles or companies are a plus