Applied Scientist

Adobe • San Jose, CA

17h

About The Position

Changing the world through digital experiences is what Adobe’s all about. We give everyone—from emerging artists to global brands—everything they need to design and deliver exceptional digital experiences! We’re passionate about empowering people to create beautiful and powerful images, videos, and apps, and transform how companies interact with customers across every screen. We’re on a mission to hire the very best and are committed to creating exceptional employee experiences where everyone is respected and has access to equal opportunity. We realize that new ideas can come from everywhere in the organization, and we know the next big idea could be yours! The Adobe Firefly Applied Science & Machine Learning (ASML) group is seeking a Model Tech Lead to own the optimization and architectural evolution of our multimodal editing foundation models. This position is central to the delivery of Firefly’s proprietary foundation models, focusing on developing efficient training and inference solutions that directly enhance our profit margins. We are looking for a hands-on technical authority with deep expertise in frontier foundation model areas to drive high-performance, large-scale AI delivery.

Requirements

• Advanced Technical Degree: Ph.D. in Computer Science, AI/ML, or a related field (or equivalent industry experience).
• Efficiency Expertise: Proven track record in model efficiency for foundation models, specifically in quantization, pruning, sparse architectures, and kernel optimization.
• Foundation Model Depth: Deep technical expertise in frontier foundation model research, including multimodal perception and generative editing.
• Leadership Impact: Experience leading multi-functional squads and defining technical roadmaps that translate complex research into production-ready, efficient models.
• Engineering Excellence: Proficiency in modern ML stacks (e.g., PyTorch, JAX) and specialized performance tools (e.g., CUDA, Triton, DeepSpeed, or FSDP).

Nice To Haves

• Experience optimizing MLLMs for high-throughput, cost-sensitive production environments.
• Familiarity with hardware-aware optimization and custom operator development.
• Background in scaling large-scale training runs while maintaining strict compute budgets and profit margin targets.
• A history of bridging the gap between ground breaking research in model sparsity and stable, enterprise-grade deployment.

Responsibilities

• Drive Architectural Innovation: Design and implement novel MLLM architectures (e.g., sparse models, MoE) and specialized kernels, using deep expertise in quantization, pruning, and post-training optimization to enhance efficiency.
• Implement Performance Roadmaps: Lead the end-to-end efficiency roadmap for Adobe’s multimodal editing stack, focusing on reducing computational costs for both large-scale training and low-latency inference while maintaining powerful performance.
• Technical Leadership & Execution: Lead a multi-functional squad as a hands-on authority, unblocking complex technical issues, conducting rigorous code reviews, and upholding high engineering standards for reproducibility and quality.
• Post-Training Optimization: Oversee advanced model compression and deployment strategies, ensuring that frontier models are optimized for real-world enterprise performance and scalability.

What This Job Offers

• Job Type: Full-time
• Career Level: Mid Level
• Education Level: Ph.D. or professional degree
• Number of Employees: 5,001-10,000 employees