Relevant Experience & Education Highlights

Deep expertise in GPU optimization, PyTorch and TensorFlow performance engineering, custom CUDA kernels, and large-scale ML model training aligns strongly with building scalable infrastructures for Adobe's Firefly generative AI foundation models.

1. You'll build and optimize infrastructures that power large foundation model training on thousands of GPUs

Demonstrated leadership in scaling ML training and inference infrastructures with GPU acceleration matches core responsibilities for foundation model stacks.

• Led ML GPU optimization efforts at a leading augmented reality and social media company, focusing on high-performance production environments.
• Improved TensorFlow and PyTorch performance on GPUs at a premier technology and AI company, optimizing models including BERT, ResNet50, and NCF with official MLPerf submissions.
• Developed custom CUDA kernels for critical bottlenecks, enhancing training and inference throughput in large-scale ML workflows.
• Engineered features and leveraged LLMs in recommendation systems at a major ads platform within the same premier company, driving model quality and revenue.

2. You will profile GPU utilization, trace inference and training runs and help craft strategies for optimizing our ML model latency

Hands-on experience with profiling tools and optimization techniques directly supports GPU utilization analysis and latency reduction strategies.

• Profiled performance using Nsight and optimized GPU memory bandwidth, kernel fusion, and model quantization (INT8/FP8) to maximize throughput.
• Deployed models with TensorRT and Triton Inference Server, slashing inference latency in production environments.
• Bridged model development and hardware by transforming computationally expensive models into efficient, low-latency applications across the ML lifecycle.

3. We'll work together to architect and optimize end-to-end ML pipelines, ensuring they're scalable, efficient, and robust

Full lifecycle experience from architecture design to deployment aligns with architecting scalable ML pipelines and systems optimizations.

• Architected novel deep learning models in PyTorch and JAX, spanning design, training, and scalable deployment.
• Applied distillation, transformers, and performance engineering techniques, including work on BERT and recommendation models with LLMs.
• Contributed to GPU modeling and simulation at a prominent consumer technology company and a major semiconductor firm, enhancing hardware utilization.

Requirements & Candidate Alignment