Staff Systems Engineer, Production-Scale AI Infrastructure

Fireworks AI; New York (NY); On-site; USD 180,000 - 300,000; Full time

Job summary

Fireworks AI seeks a Member of Technical Staff for the Systems Infrastructure team to design and build the substrate that powers thousands of accelerators. You will tackle scheduling, storage, and networking challenges in a production fleet, with mentorship from senior engineers and project ownership from day one.

PhD graduates or soon-to-be PhDs in relevant fields will work on real problems, delivering measurable improvements in latency and throughput across heterogeneous hardware environments.

Qualifications

• PhD completed or expected by Dec 2026 in CS/CE/EE or related field.
• Strong research background in distributed systems, OS, scheduling, storage, networks, or HPC.
• Depth in at least one focus area demonstrated via dissertation, publications, or completed systems.
• Proficient in C/C++, Rust, Go, and Python; capable of real systems development.
• Experience building and evaluating real systems with rigorous measurements.
• Ability to communicate systems designs and results clearly to diverse audiences.

Responsibilities

• Design, build, and operate core infrastructure systems for large-scale training and inference.
• Model and measure system behavior; build benchmarks, traces, and simulators to reason about scheduling, caching, and networking.
• Identify bottlenecks across the stack and quantify improvements with data.
• Turn research ideas into production systems that handle real workloads and failures.
• Collaborate with research and inference teams so infrastructure and model design inform each other.
• Contribute to technical direction by tracking emerging hardware and interconnects.

Skills

C/C++

Rust

Go

Python

Systems programming

Benchmarks & measurement

Education

PhD in CS/CE/EE or related

Tools

Kubernetes

Ray

Slurm

NCCL

DPDK/SPDK

Ceph

Job description

Fireworks AI seeks a Member of Technical Staff for the Systems Infrastructure team to design and build the substrate that powers thousands of accelerators. You will tackle scheduling, storage, and networking challenges in a production fleet, with mentorship from senior engineers and project ownership from day one.

PhD graduates or soon-to-be PhDs in relevant fields will work on real problems, delivering measurable improvements in latency and throughput across heterogeneous hardware environments.