SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Inferact, founded by the creators and core maintainers of vLLM, is seeking an inference runtime engineer to advance the state of LLM and diffusion model serving. You'll work at the core of vLLM, optimizing how models execute across diverse hardware and architectures. As models grow larger and architectures evolve—mixture-of-experts, multimodal, agentic systems—the inference engine itself must innovate to keep pace. Your work will directly impact how the world runs AI inference, making it cheaper and faster.
You'll tackle complex challenges in transformer architecture optimization, KV-cache memory management, prefix caching, and hybrid model serving. The role demands deep expertise in LLM inference systems, the ability to read and implement techniques from research papers, and the skill to contribute performant, maintainable code to complex ML codebases.
Minimum qualifications include a bachelor's degree in computer science or equivalent, deep understanding of transformer architectures and variants, strong Python and PyTorch internals experience, and hands-on experience with LLM inference systems like vLLM, TensorRT-LLM, SGLang, or TGI. You should be able to debug in complex ML codebases and translate research into production code.
Preferred experience includes deep knowledge of KV-cache memory management and prefix caching, familiarity with RL frameworks for LLMs, multimodal inference work (audio/image/video/text), and contributions to open-source ML or system infrastructure projects. Bonus points for core vLLM contributions, integrations with projects like verl or OpenRLHF, or published technical content on LLM inference.
The role is fully remote and timezone-flexible, though regular overlap with Pacific Time is expected for critical syncs. Inferact offers competitive salary and equity packages calibrated to local market conditions, visa sponsorship on a case-by-case basis, and location-appropriate benefits including health coverage.
