Performance & Debug Engineer for AI Inference

Positron AI; United States; Remote; USD 200,000 - 300,000; Full time

Benefits offered by this job

Fully company-paid health, dental, and

Vision insurance

Unlimited paid time off

401(k) with company matching

Equity

Job summary

Positron AI is hiring a Performance & Debug Software Engineer for AI inference on custom accelerators. You will optimize end-to-end LLM inference performance across FPGA Atlas and Titan systems, profiling hardware-accelerated attention, data movement, and runtime behavior.

You will lead benchmarking, regression testing, and cross-layer debugging from application to hardware, with opportunities to influence architecture and interfaces.

Qualifications

• Strong C++ and Python skills with hands-on profiling/optimization on complex systems involving Linux VFIO.
• Deep understanding of memory hierarchy, concurrency, data movement, batching, scheduling, and performance measurement.
• Hands-on experience with profilers, counters, traces, and statistical methods to separate signal from noise.
• Ability to debug across application, runtime, driver, firmware, and hardware boundaries.
• Experience bringing up software on new silicon and debugging performance optimizations in production deployments.
• Enthusiasm for AI-accelerated software development and coordinating subagents to increase speed without compromising quality.

Responsibilities

• Profile and optimize latency, throughput, utilization, and efficiency across hardware-accelerated attention and device runtimes.
• Use performance counters, traces, telemetry, and simulation to isolate regressions and validate fixes.
• Develop and optimize distributed inference capabilities, incl. efficient KV cache management and data movement.
• Build benchmarks and workload traces to expose bottlenecks in compute, memory, interconnect, networking, and scheduling.
• Partner with test engineers to turn findings into regression tests and release gates.
• Influence hardware/software architecture and interfaces based on measured performance.

Skills

C++

Python

Memory & Concurrency

Profiling & Tracing

Cross-layer Debugging

Silicon Bringup

AI acceleration

Tools

CUDA

HIP

Triton

GPU kernel development

Distributed inference

KV cache systems

Job description

Positron AI is hiring a Performance & Debug Software Engineer for AI inference on custom accelerators. You will optimize end-to-end LLM inference performance across FPGA Atlas and Titan systems, profiling hardware-accelerated attention, data movement, and runtime behavior.

You will lead benchmarking, regression testing, and cross-layer debugging from application to hardware, with opportunities to influence architecture and interfaces.