Director of ML Engineering - Conversation Product Area
Spotify
Job Description
The Personalization team makes deciding what to play next easier and more enjoyable for every listener. From Blend to Discover Weekly, we're behind some of Spotify's most-loved features. We built them by understanding the world of music and podcasts better than anyone else. Join us and you'll keep millions of users listening by making great recommendations to each and every one of them.
Spotify's Personalization mission (PZN) is building the single intelligence layer that powers how hundreds of millions of people discover and experience audio. One agent, one reasoning engine, and one taste model, moving to a unified single stack.
We're looking for a Director of Engineering to lead the Conversation Product Area. The Conversation Product Area is the technical center of gravity for the shift to one Agent, building the shared Agent Engine harness and the evaluation infrastructure that supports it. The Product Area also owns the Search Brain (intent routing and search-driven conversation) the intent-detection and prompt-suggestion platform behind our agentic consumer features like Chat and AI DJ, and the Speak Platform, Spotify's first-party voice team, which spans applied research, production infrastructure, and the character/brand decisions behind every voice Spotify users hear. This is the engineering org where architecture decisions get made that power every agentic surface at the company.
You'll be working with many of Spotify's senior architects, with product leadership in the consumer AI space, and with cross functional partners. The team needs a leader who is as comfortable making critical architectural calls as they are managing the people making them.
What You'll Do
- Own engineering execution and technical direction for the Conversation Product Area, including the Agent Engine, the evaluation infrastructure, the Search platform, the AI Voice platform (Speak), and some consumer agentic features built on top of the Agent Engine. This spans production systems at massive scale, applied ML research, and infrastructure that other product areas depend on.
- Co-lead the Product Area in a paired product/engineering model. You own engineering and deeply partner with a product counterpart that owns product strategy. You’ll operate as a single leadership unit for planning, prioritization, and external representation. This partnership is core to the success of the role.
- You’ll manage a team of ~70 engineers, with a small team of Senior Engineering Managers and Staff/Principal engineers directly reporting to you. You’ll develop and hold accountable the EM layer running the PA's roughly 9 teams. Your leverage comes through your SEMs and Staff engineers, but you’re ultimately accountable for delivery across the entire product area.
- Drive the migration of the company’s existing agentic infrastructure onto the shared Agent Engine, in close partnership product teams across the company, balancing migration speed against product teams’ existing roadmap work.
- Own the technical strategy for Spotify's first-party voice investment: what Speak builds versus buys, how the team organizes and delivers against a market that's commoditizing fast, and how the character/brand voice decisions for agent experiences get made and shipped on their own aggressive timelines.
- Define and hold the technical bar on evaluation. Build the discipline around building feature-specific judges and datasets and product decisions based on those eval sets.
- Set and enforce engineering quality standards appropriate to an AI-assisted development era including code review discipline and test coverage especially designed for teams that are increasingly writing code with AI assistance.
- Participate as a full member of the broader leadership group, co-developing with your peers and executives the operating model for the whole studio across capacity management, project prioritization, studio communication principles, and building a healthy engineering culture in a highly changeable environment.
- Represent the Conversation PA's engineering point of view in high visibility, company-wide architecture and steerco conversations with senior leaders.
Who You Are
- You are deeply technical, not tech-adjacent. This PA's defining work (a new shared agent runtime and harness, a scaled evaluation system, a voice/speech stack) requires a Director who can evaluate architecture proposals, hold their own in a room with senior ICs, contribute technically in meaningful ways, and make tough tiebreaking and tradeoff calls.
- You have run large engineering orgs through platform consolidations, migrations, or evolutions, and you’ve had to earn trust by building highly scaled and functional platform technology.
- You have experience with at least two of: LLM-based agent systems and orchestration, evaluation/eval-infrastructure design (including automated judges and regression suites), large-scale retrieval or search systems, or speech/voice ML systems and production voice infrastructure.
- You've managed managers, not just individual contributors. You know how to build a management structure that scales.
- You have a track record of building strong but nimble teams, with experience restructuring, shifting team resourcing, adjusting to fast changing priorities or shifting needs without losing delivery momentum.
Where You'll Be
- We offer you the flexibility to work where you work best! For this role, you can be within the North Americas or EMEA region as long as we have a work location.
- This team operates primarily within the Eastern North America and Central European time zones for collaboration, so we ask that you be located within a time zone that enables regular overlap with the team.