Candidate Profile — Candidate #2
Position: Adobe — Machine Learning Infrastructure Engineer
Candidate Location: San Francisco Bay Area
Experience: 16.8 years experience
Relevant Experience & Education Highlights
Seasoned AI infrastructure engineer with deep expertise in GPU/TPU optimizations, custom CUDA kernels, and scalable ML pipelines strongly aligns with building high-impact foundation model infrastructures for Adobe's Firefly generative AI efforts.
1. Build and optimize infrastructures that power large foundation model training on thousands of GPUs
Engineered high-performance ML systems leveraging GPU and TPU accelerations for demanding workloads.
- Developed GPU-based ML engines with custom CUDA kernels for algorithms including k-means, PageRank, GEMM, and GEMV at a premier research center.
- Optimized TensorFlow models for performance and power on tensor processing chips at a major technology company.
- Built high-performance graph analytics engines exceeding GraphLab and GraphX efficiency by over 1,000x in performance per watt for IoT applications.
- Shipped ML applications including realtime camera processing, speech recognition, and low-light imaging on mobile hardware platforms.
2. Profile GPU utilization, trace inference and training runs and help craft strategies for optimizing our ML model latency
Profiled and traced ML inference and training to deliver power-efficient, low-latency solutions across edge and cloud-scale environments.
- Profiled and optimized GPU/TPU utilization for CNNs like MobileNet, Inception, ResNet, and RNNs like LSTMs with quantization at a leading tech firm.
- Implemented fixed-point Fast Fourier Transform in C++ on Pixel Visual Core and Edge TPU for face detection and HDR+ burst photography.
- Engineered performance and power optimizations for TensorFlow models in FaceAuth, OCR, speech recognition, and Top Shot video applications.
3. Architect and optimize end-to-end ML pipelines, ensuring they're scalable, efficient, and robust
Architected scalable ML pipelines from data processing to deployment, emphasizing efficiency and reliability.
- Led development of parallel graph analytics, matrix factorization, and recommender systems on GPU/TPU platforms.
- Integrated continuous regression and model-based control for ML applications in embedded reasoning and planning systems.
- Deployed ML platforms supporting realtime apps on ARM/x86 processors with veteran expertise in systems building and product shipping.
Requirements & Candidate Alignment
| Adobe Requirement | Candidate Qualification |
|---|---|
| Education: Graduate, PhD, or postgraduate degree in Computer Science, Computer Engineering, or a related field—or equivalent experience. | PhD in Artificial Intelligence from a research university, plus Master's in Robotics from a leading engineering university. |
| Experience: 2+ years ML Engineering experience, specializing in generative AI like LLMs. | 16.8 years of ML engineering experience across GPU/TPU platforms, deep learning, and AI systems. |
| Programming and Frameworks: Strong Python and deep learning engineering skills, paired with experience in training and inferencing with PyTorch or TensorFlow, will be essential. | TensorFlow training and inference expertise with CNNs, RNNs, and quantization techniques. |
| Model Familiarity: Familiarity with distillation, transformers, and diffusion models. Experience with generative image and video is a plus. | Quantization for model optimization aligning with distillation; hands-on generative imaging via HDR+ and Top Shot video, plus OCR and speech recognition. |
| Deployment Technologies: Knowledge of deployment technologies such as Docker, ML Ops, and ML services is valuable. | ML platform deployment for edge TPU and visual core hardware, including realtime inference services. |
| Cloud Platforms: experience with cloud platforms like Azure and AWS is a plus. | Cloud infrastructure experience supporting scalable ML workloads and big data analytics. |
| Problem-Solving: excellent problem-solving abilities and your capacity to analyze complex issues and drive solutions with a data-driven approach. | Data-driven performance optimizations for GPU/TPU ML, graph analytics, and embedded systems. |
| Communication: strong verbal and written communication skills and success in cross-functional team environments. | Led cross-functional teams with 100+ publications and editorial board role at a leading AI journal. |
If you would like to discuss this candidate or other critical roles, here is a link to my calendar to schedule a call:
Schedule a CallContact:
Finding the signal in the noise since 2005