Lead Inference Engineer (Fully Remote)

Aleroy; United States; À distance; USD 140 000 - 190 000; Plein temps

Résumé du poste

Aleroy seeks a senior engineer to own and enhance the AI inference serving stack in the United States. You will drive performance through CUDA and Triton, implement efficient batching and cache strategies, and mentor engineers while staying hands-on with critical implementations.

You will help shape the roadmap for serving and measurement, establish engineering standards, and push for first-class reliability and scalability across inference workloads.

Qualifications

• Own and improve the AI inference serving stack (prefill/decode disaggregation, batching, KV-cache, speculative decoding).
• Identify and eliminate GPU bottlenecks using CUDA, Triton, profiling, and kernel optimization.
• Set the serving/measurement roadmap, establish engineering standards, and mentor engineers while remaining hands-on.

Connaissances

GPU performance engineering

CUDA

Triton

Profiling

Kernel optimization

Mentorship

Hands-on implementation

Outils

Triton Inference Server

CUDA Profiling

Description du poste

• Own and improve the AI inference serving stack, including prefill/decode disaggregation, continuous batching, KV-cache management, and speculative decoding.
• Drive GPU performance engineering by identifying and eliminating bottlenecks using CUDA, Triton, profiling, and kernel optimization.
• Set the serving and measurement roadmap, establish engineering standards, and mentor engineers while remaining hands-on with implementation.