Director, ML Engineering

Adobe•San Jose, CA

7d•Remote

About The Position

Firefly Foundry is a new business venture at Adobe, offering an enterprise-managed service for custom multimedia generative AI. This service includes custom image, video, and 3D models tuned to customer IP, integrated with creative workflows for content production and VFX, deployed across Adobe surfaces, and supported by a media intelligence layer. The business has seen significant growth in Media & Entertainment, marketing, and consumer retail, and is expanding into new sectors. This role is for a Director, ML Engineering who will lead the engineering function for Firefly Foundry's model services at an enterprise scale. It's a comprehensive executive position with full accountability for the production, serving, and operation of Firefly Foundry's models for enterprise clients, as well as for the engineering organization responsible for this capability. The role involves defining and executing the technical strategy to scale Firefly Foundry from a managed service for early partners to a platform serving hundreds of enterprise customers concurrently, each with unique IP, tenancy, and SLAs. This includes powering franchise extensions, new IP development, and modern GenAI workflows at production scale.

Requirements

10+ years in applied machine learning and ML systems.
5+ years leading engineering organizations, with prior experience leading managers of managers.
Demonstrated success shipping generative AI products in production at enterprise scale.
Proven ability to operate as a peer to VP-level partners across product, science, infrastructure, and field organizations.
Ability to represent engineering credibly in front of senior customer and partner executives.
Track record of building engineering benches, defining career frameworks for emerging roles, and developing leaders.
Deep understanding of the modern generative model landscape (diffusion, transformers, VAEs, latent video models, control/adapters, or similar).
Strong intuition for the economics and engineering reality of large-scale inference: accelerator stacks, model optimization and quantization, and tradeoffs between quality, latency, and cost.
Experience designing and operating ML systems end-to-end (data, training, evaluation, deployment, monitoring, continuous improvement) at production scale.
Executive presence and communication skills.
Ability to brief executives, customers, and partners credibly.
Ability to work directly with creative, production, and business collaborators to turn ambiguous problems into clear technical plans.
Sound judgment under ambiguity, comfortable making decisions with incomplete information and revising as new information arrives.
MS or PhD in Computer Science, Electrical Engineering, or a related field, or equivalent practical experience building and leading advanced ML systems.

Nice To Haves

Familiarity with high-resolution media pipelines (4K+ video, high bit-depth) or adjacent bandwidth- and latency-sensitive domains is a meaningful plus.

Responsibilities

Define and implement the technical strategy for productionizing custom generative models across image, video, and 3D, including architectural patterns for simplified pipeline construction and efficient absorption of diverse internal and external models.
Own the unit economics of Firefly Foundry inference, including cost-to-serve, GPU utilization, and gross margin.
Define the tenancy and data-isolation architecture to ensure compliance with enterprise IP contracts.
Drive the self-serve roadmap to expand Firefly Foundry's reach and value beyond direct engagements.
Represent Adobe engineering in C-suite and senior technical discussions with studios, brands, and global enterprises.
Build and lead a multi-team engineering organization of ML engineers and engineering managers, focusing on recruitment, hiring, development, and retention of senior technical and leadership talent.
Establish the engineering bar, bench, and talent strategy to support 10x growth in capability breadth and traffic without linear headcount growth.
Define the operating rhythm, including goal-setting, executive reviews, and engineering reviews, to ensure coordination in a fast-scaling organization.
Own the multi-year architecture for training and inference at scale, covering pipeline construction, data pipelines, evaluation frameworks, model lifecycle management, and accelerator utilization.
Set the strategy for fast model deployment, parallel pipeline operation, tenancy/data isolation, and self-serve capability development.
Make build-vs-buy and prioritization decisions on emerging GenAI techniques in collaboration with Applied Science.
Own production reliability and economics, ensuring production SLAs for model services.
Oversee analytics and observability across all model pipelines, monitoring quality, latency, cost, and utilization.
Drive down cost-to-serve over a multi-year period while expanding capability.
Represent engineering in technical customer engagements, translating requirements into ML roadmaps and success metrics.
Co-design scalable, cost-efficient serving for real-time and high-volume content generation.
Steward relationships with GPU vendors and hyperscalers to secure serving capacity.