Inference Performance Engineer | Flexible, High-Impact ML

adaption; Argentina; Híbrido; ARS 2.000.000 - 4.200.000; Jornada completa

Ventajas ofrecidas por este puesto de trabajo

Flexible work

Travel stipend

Lunch stipend

Well-being

Descripción de la vacante

Adaption is seeking a seasoned ML systems engineer to own the cost and performance of its inference stack. You will shape throughput and tail latency by tuning caching, batching, quantization, and decoding while preserving model quality.

You’ll collaborate with the serving fleet team, optimize routing to providers, and build profiling tools to reveal time, memory, and compute usage. Ideal candidates have 5+ years in ML systems and hands-on experience with vLLM, SGLang, or TensorRT-LLM, plus

Formación

• 5+ years of ML systems or inference infrastructure experience with measurable improvements.
• Deep understanding of model serving, including prefill and decode, memory bandwidth, batching, and concurrency.
• Production experience with serving engines such as vLLM, SGLang, or TensorRT-LLM.
• Strong Python skills and proficiency in C++, Rust, or another systems language.

Responsabilidades

• Improve throughput, cost, and tail latency through KV-cache management, continuous batching, speculative decoding, and quantization.
• Optimize long-context prefill and decode workloads based on real production traffic.
• Tune routing between our infrastructure and external providers based on cost, capacity, and performance.
• Work within serving engines such as vLLM, SGLang, and TensorRT-LLM, going below the framework when needed.
• Build profiling and measurement systems that show where time, memory, and compute are being spent.

Conocimientos

5+ years in ML systems

Model serving/inference

Python

C++

Rust

CUDA/NCCL/mixed precision

GPU performance

Herramientas

vLLM

SGLang

TensorRT-LLM

Descripción del empleo

Adaption is seeking a seasoned ML systems engineer to own the cost and performance of its inference stack. You will shape throughput and tail latency by tuning caching, batching, quantization, and decoding while preserving model quality.

You’ll collaborate with the serving fleet team, optimize routing to providers, and build profiling tools to reveal time, memory, and compute usage. Ideal candidates have 5+ years in ML systems and hands-on experience with vLLM, SGLang, or TensorRT-LLM, plus