Sr Staff Machine Learning Engineer - Media Intelligence

Adobe•San Jose, CA

31d•$190,200 - $345,650

About The Position

Firefly Foundry is Adobe's enterprise managed-service offering for custom multimedia generative AI — deep-tuned image, video, and 3D models built on each customer's IP, paired with creative production workflows and a media-intelligence layer, and deployed across new and existing Adobe surfaces. As customers bring ever-larger libraries of media and generate ever more of it, the ability to process, understand, and search that content — for people and, increasingly, for AI agents — is becoming one of the product's core forms of leverage. The business has gained significant traction in Media & Entertainment, marketing, and consumer retail, and is expanding rapidly into adjacent verticals. We are hiring a Senior Staff Machine Learning Engineer to architect and lead the data processing, indexing, and search infrastructure behind Firefly Foundry's media intelligence — the systems that turn massive volumes of customer media (image, video, 3D, audio) and model-derived signals (embeddings, captions, entities, shot and scene structure, aesthetic, safety, and IP labels) into structured, low-latency, searchable intelligence, and that expose it as agentic search: retrieval designed to be driven by AI agents, not only by people. This is a systems and infrastructure role, not a model-training or research role — you won't be running training experiments. You own the platform on the other side of the model: the pipelines that enrich and index media at scale, the hybrid and multimodal retrieval stack that serves it, and the tool interfaces and grounding contracts that let agentic workflows retrieve, reason, and cite. As a Senior Staff engineer you set the multi-year technical direction for this platform, are the recognized technical authority for data and search across Firefly Foundry, and multiply the teams around you through design leadership and mentorship. Your work has direct, measurable impact on the recall, freshness, latency, cost, and scale of everything Firefly Foundry's intelligence and agents depend on.

Requirements

10+ years in machine learning, data, or infrastructure engineering, including deep ownership of large-scale data processing and/or search & retrieval systems in production — and a track record of leading systems and setting technical direction across teams.
Deep expertise designing and operating search and retrieval infrastructure at scale — vector/ANN (e.g., HNSW, IVF, ScaNN, DiskANN), lexical search (Lucene / Elasticsearch / OpenSearch), hybrid retrieval, ranking and reranking, and query understanding.
Strong data-engineering foundations — large-scale batch and streaming pipelines (e.g., Spark, Beam, Flink, Ray), data modeling, and the storage systems behind them (object stores, vector databases, columnar/OLAP).
Experience building retrieval for LLM and agentic systems — RAG, multimodal and cross-modal search, grounding and provenance, and retrieval evaluation.
Strong Python, plus a systems language (Go, Rust, or C++) a plus; hands-on familiarity with embedding models and the inference paths that produce them (PyTorch).
A track record building the observability, monitoring, and alerting that data and search systems rely on to hit freshness, recall, and latency SLAs.
Experience with multi-tenant systems and data isolation in an enterprise or regulated context.
Fluency with containers and orchestration (Docker, Kubernetes), CI/CD, and a major cloud (AWS or Azure).
Comfort reasoning about retrieval quality and relevance across modalities (text, image, video, 3D, audio), in partnership with Applied Science.
Proven technical leadership — mentoring senior engineers, driving cross-org design and build/buy decisions, and influencing roadmap and standards.
Excellent communication and data-driven problem-solving in cross-functional settings, including with leadership.

Responsibilities

Design and build scalable data-processing pipelines that transform raw customer media and model-derived signals (embeddings, captions, entities, shot/scene structure, safety and IP labels) into structured, searchable intelligence — with the throughput, correctness, and cost profile enterprise scale demands.
Contribute to the technical vision and architecture for Firefly Foundry's media-intelligence data platform and search stack — the systems that ingest, enrich, index, and serve retrieval over billions of media assets — and be the engineer the organization looks to for the hardest data and search decisions.
Architect the indexing and search infrastructure — hybrid lexical + vector (ANN) retrieval, multimodal and cross-modal search, ranking and reranking, faceting and rich metadata filtering — tuned for both human and agent consumers.
Make search a first-class capability for agents — tool/function-call retrieval interfaces, multi-hop query planning, iterative retrieval, and grounded results with citations and provenance that agentic workflows can trust.
Own index lifecycle and freshness — incremental and streaming indexing, backfills and reprocessing, and schema and embedding-model versioning — so the index stays correct and current as models and content evolve.
Engineer for enterprise from the ground up — per-tenant index isolation, data residency, and the access controls that let us honor customer IP contracts under audit.
Define and enforce retrieval quality gates — offline and online evaluation (recall@k, nDCG, groundedness), regression detection, and drift monitoring — that block quality regressions from reaching production.
Own the performance and cost envelope of the platform — query latency (p50/p99) and throughput SLAs, ANN index tuning, GPU-accelerated enrichment (embedding/captioning) at scale, and right-sizing storage, serving, and accelerator fleets.
Build the platform underneath it all — rapid pipeline and index deployment, observability, monitoring, and alerting across data and search systems.
Run these systems operationally at enterprise scale — on-call, incident response, and postmortems for availability, freshness, and latency regressions.
Lead technically across teams — set standards, drive build/buy and design decisions, mentor senior engineers, and represent Firefly Foundry's data and search architecture to leadership and partner orgs.