Software Engineer, AI Engine / HPC

Topaz Labs (Headquarters: Dallas, TX)

Software Engineer C++ HPC High Performance Computing Performance Optimization GPU CUDA Metal AI Engine Inference Deep Learning Systems Programming Python AI Engineer Staff Engineer

Company: Topaz Labs is a rapidly growing AI software company specializing in photo and video enhancement, bootstrapped to $48M revenue. We build tools used by professionals and creatives worldwide, including NASA, Nvidia, Apple, Google, and Tesla. We have over 1.5 million users and foster a culture of craftsmanship, ownership, and fast iteration in our Dallas, TX headquarters.

Role: Software Engineer, AI Engine / HPC

Responsibilities: Develop, optimize, and maintain the core AI inference engine that powers Topaz Labs products. Focus on high-performance computing (HPC), optimizing model execution on various hardware (CPU, GPU, NPU) using low-level APIs (CUDA, Metal, DirectML, OpenVINO etc.). Implement efficient data pipelines and memory management strategies. Profile and benchmark performance. Work closely with research scientists to integrate and optimize new AI models. Primarily using C++ and Python.

Skills Required: Strong C++ programming skills and experience with systems-level development. Experience with high-performance computing (HPC) and performance optimization techniques. Familiarity with GPU programming (CUDA, Metal, OpenCL) or other hardware accelerators. Understanding of deep learning inference frameworks and model optimization (e.g., TensorRT, ONNX Runtime). Strong problem-solving and debugging skills. Python proficiency is also required.

Ideal Candidate: A performance-obsessed software engineer with expertise in C++ and HPC. Enjoys optimizing code for speed and efficiency on diverse hardware platforms. Interested in the intersection of AI/ML and systems programming. Excited to build the core engine powering cutting-edge AI applications.

Compensation: Competitive salary (Range: $130,000 - $250,000, commensurate with experience) and equity package.