Engineering Manager, Agent

About the Role

Descript is revolutionizing how traditional editing methodologies and AI technologies come together in every part of the creation process (recording, transcribing, editing, and sharing) to make it faster and easier than ever to create high quality content.

We are hiring an Engineering Manager to lead our Agent team, focused on building the world’s best Agentic Video Editor, Underlord. This is a rapidly changing landscape, and there are many unique technical challenges that come with building high quality, cost-effective Agents for handling the multimodal nature of video. You will manage a team of 5-6 software engineers, report to the Head of AI Engineering, and partner closely with others across engineering, product, design, and AI research teams.

Agent-driven editing is a major company focus and this role requires strong product thinking and strategic partnership skills as the team continues to develop novel, high-quality experiences around Agentic Video Editing.

About the team:

Descript's Agent launched in Open Beta in July 2025. Since then, we've been working on improving the user experience for chat-based editing, adding more capabilities into our agent, and improving our agent infrastructure to continue to scale as models get better and better. The Agent product engineering team partners very closely with our AI Research team, in order to achieve our mission to make video editing accessible to anyone. Your management peers and direct reports will be based in San Francisco, New York, and distributed across the US and Canada.

This is the role for you if you are excited about driving progress in the following areas:

• Building a best-in-class product experience for agentic video editing interactions (eg. tuning prompt templates and other key user workflows, addressing user feedback, building key product capabilities like chat history, etc)
• Driving rapid, data-driven experimentation to push the limits on quality for the video editing agent (eg. model/routing experimentation, pricing experimentation, core agent investments around harness and tool design and token optimizations, etc)
• Deepening the Agent Quality flywheel to better prioritize our roadmap by leveraging user data and signals to strengthen Evals and product analysis methodologies
• Partnering with AI Research on roadmap for exploratory Agent improvements (eg. novel context management strategies, RL/new model development, multimodel explorations, etc)

What you'll do:

Your primary focus in 2026 will be driving quality to move from beta to GA and delivering material impact for the business. In this role, you will:

• Own execution: Manage team prioritizes, drive alignment across stakeholders, adapt plans to maximize outcomes, and hold a high bar for delivery through active planning, prioritization, and follow-through
• Define team direction: Connect customer needs, product vision, and technical realities in close collaboration with Product, Design, and AI Research by setting clear priorities and making tradeoffs to deliver results
• Develop your team: Recruit, mentor, and grow engineers, fostering a culture of continuous learning and high performance. Create stretch opportunities for ICs, deliver clear and direct feedback, and manage underperformance early
• Drive operational excellence: Ensure execution is predictable, reliable, and sustainable, implementing best practices in project management and engineering processes
• Maintain the technical bar: Guide architectural decisions, ensure sound design tradeoffs, address tech debt that slows execution, and drive delivery of high-quality, reliable systems

What You Bring

• Minimum Qualifications
  • Engineering Leadership Experience: 5+ years of software engineering experience, including 2+ years of engineering management experience delivering complex, cross-functional product work with high velocity and quality
  • Team Builder: Excels at hiring, developing, and retaining strong engineering talent, and creating collaborative, empowering, and high-performing team environments
  • Technical Judgment: Stro