GPU/FPGA Software Engineer

Location: Annapolis Junction, Maryland, US; onsite role.

Salary: $60.70–$105.76 per hour.

Company

EverWatch provides software, defense, intelligence, and deployed support solutions for critical US Government missions.

What you will do

• Architect and implement high-throughput, low-latency acceleration pipelines using FPGAs or GPUs.
• Develop efficient software from concept through production-ready systems using C++, Python, Docker, APIs, analytics, and infrastructure.
• Analyze current architectures and operating environments to support future functionality and enhancements.
• Deliver complete end-to-end solutions using modern architectural approaches, open-source frameworks, and tools.
• Identify and implement system improvements in an agile environment.

Requirements

• TS/SCI clearance with a polygraph is required.
• Bachelor’s degree and 20+ years of software engineering experience, or 24+ years of software engineering experience in lieu of a degree.
• Experience architecting and implementing high-throughput, low-latency acceleration pipelines with FPGAs using RTL/HLS or GPUs using CUDA/HIP.
• Experience with C++, Python, and Docker.
• Understanding of GPU microarchitecture, including warp/wavefront schedulers, shared memory utilization, and occupancy optimization.
• Experience with SystemVerilog, Verilog, or VHDL for FPGA development, or C/C++ for HLS.
• Experience with AMD/Xilinx Vivado, Vitis, or Intel Quartus Prime.
• Experience integrating high-speed IPs such as PCIe Gen 4/5, DDR4/5, HBM, or 100G Ethernet.

Nice to have

• Experience with Agentic AI or machine learning.
• Experience with Kubernetes or Docker Swarm.
• Experience with Agile.

Culture & Benefits

• Work in an exciting, agile environment focused on difficult US Government mission challenges.
• Use modern architectural approaches, open-source frameworks, and tools.
• Competitive salary and benefits package.
• Employment is subject to a security investigation and eligibility requirements for classified information access.