Senior ML Engineer | Kimchi (Inference Stack Optimization)

Kimchi is hiring a Senior ML Engineer to optimize the inference stack behind its serverless and customer-hosted model serving. The role focuses on improving throughput, latency, cold starts, and cost per token across large checkpoints and changing traffic patterns. You will work with serving engines such as vLLM and SGLang, GPU systems, storage paths, quantization, and multi-node deployments.

You will lead optimization work from measurement through implementation, using benchmarks and reproducible experiments to guide technical choices. Responsibilities include tuning schedulers and KV caches, reducing checkpoint load times, profiling latency, scaling distributed inference, and setting the technical direction for the stack. The team seeks 5+ years building ML systems and hands-on experience with inference or training infrastructure. The position is remote-first, with eligibility explicitly listing Italy.

Highlights

• Optimize production inference performance and cost.
• Work across vLLM, SGLang, GPU systems, and distributed serving.
• Own benchmark strategy and technical direction.
• Equity options and a learning budget.