Inference Performance Engineer: Optimize ML Serving & Latency (Flexible Work)

adaption; Dublin; On-site; EUR 120,000 - 180,000; Full time

Benefits offered by this job

Flexible work

Adaption Passport

Lunch stipend

Well-being benefits

Job summary

Adaption is seeking an senior ML systems engineer to own the cost and performance of the inference stack. You will optimize caching, batching, and decoding while coordinating with the serving fleet to improve throughput and latency without compromising model quality.

You will work with engines like vLLM, SGLang, and TensorRT-LLM, implement profiling tools, and help route workloads to balance cost and performance in a global deployment.

Qualifications

• 5+ years in ML systems, inference infrastructure, or performance engineering with measurable improvements in cost or latency.
• 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.
• Experience with GPU performance, including CUDA, NCCL, mixed precision, memory layout, kernels, or quantization.

Responsibilities

• 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.

Skills

Python

C++

Rust

Performance engineering

Tools

vLLM

SGLang

TensorRT-LLM

CUDA

NCCL

Job description

Adaption is seeking an senior ML systems engineer to own the cost and performance of the inference stack. You will optimize caching, batching, and decoding while coordinating with the serving fleet to improve throughput and latency without compromising model quality.

You will work with engines like vLLM, SGLang, and TensorRT-LLM, implement profiling tools, and help route workloads to balance cost and performance in a global deployment.