Director of Research, Agentic AI

DigitalOcean is building the Agentic Inference Cloud — anchored by Inference Engine, our model-serving platform — to help fast-growing AI-native companies start, scale, and optimize agentic workloads. Applied Research is the layer that generates the proprietary science that compounds across that platform: routing that learns instead of following static rules, memory that improves recall and personalization, observability that explains why agents succeed or get stuck, and reinforcement learning that closes the loop between production signals and model behavior.

We're hiring a Director of Research to build and lead this function: set the research agenda, grow the team, and make sure the science this group produces ships into real product outcomes — better model selection and routing, more reliable agents, lower cost and latency, and higher task-success rates for the developers and enterprises running agentic workloads on DigitalOcean.

This is a hybrid role by design: enough technical depth to personally evaluate and shape research direction across routing, memory, observability, and RL, and enough leadership range to build a team, prioritize against a roadmap, and defend research investment to product and executive stakeholders.

What You'll Do

Set the research agenda

• Define and own the applied research roadmap across adaptive routing (evolving model selection from static rules into a system that learns from real usage, cost, and latency), memory (retrieval quality and durable recall for long-running agents), agent observability (understanding when agents make progress, get stuck, or make mistakes), and reinforcement learning / closed-loop learning (turning production feedback into better models and policies).
• Track emerging model architectures and specialized, domain-tuned model approaches, and translate what's relevant into DigitalOcean's product roadmap.
• Keep the agenda tightly coupled to product outcomes — every research bet should map to a measurable improvement in model selection, agent reliability, cost/latency, or task-success rate, not research for its own sake.

Build and lead the team

• Grow the Applied Research team, hiring and mentoring research scientists and engineers.
• Establish the team's operating rhythm: how research questions get scoped, how experiments get run and evaluated, and how findings hand off to production teams.
• Represent Applied Research in cross-functional planning cycles alongside other engineering and product leaders, and make the case for headcount and investment on its own merits.

Bridge research and product

• Partner directly with Inference Engine and the other platform and product engineering teams that own agent runtime and evaluation infrastructure to turn research into shipped capability.
• Turn research prototypes into production-ready capabilities in partnership with engineering — shipping research, not just publishing it.
• Communicate research trade-offs clearly to non-research stakeholders, including when a promising direction isn't ready for product investment yet.

Represent DigitalOcean externally

• Build DigitalOcean's credibility in the applied agentic-AI research community through publications, talks, open-source contributions, or collaborations — where they serve product and hiring goals.

What You'll Add to DigitalOcean

• 10+ years in applied ML/AI research or research-adjacent engineering, including experience leading a research team or function — formal people management or clear de facto technical leadership of a research group.
• Deep, hands-on expertise in at least two of: LLM routing and model selection, retrieval and memory systems, agent observability and evaluation, or reinforcement learning (RLHF, RLAIF, DPO, PPO, GRPO, or related methods).
• A track record of shipping research into production systems — not just publishing or prototyping it.
• Fluency with the current agentic AI landscape: reasoning, planning, tool use, long-horizon memory, and the practical economics of large-scale model serving.
• Strong technical communication — able to defend a research agenda to engineering leaders, product leaders, and executives, and to translate ambiguous research questions into a roadmap with concrete checkpoints.

Preferred Qualifications

Strong signal

• Experience building or scaling an applied research team inside a product organization (not a pure research lab), with research investment justified in business terms.
• Direct experience with inference infrastructure or large-scale model-serving optimization.
• Familiarity with the broader open-weight and open-source model ecosystem.

Nice to have

• PhD in CS, ML, or a related field — or equivalent depth demonstrated through industry impact.
• Publications, patents, or open-source contributions in routing, memory, agent evaluation, or reinforcement learning.
• Experience with small, specialized models or domain-tuned LLMs.

Compensation Range: 

• $249,600 - $312,000

*This is a hybrid role

JR: 2026-8125

#LI-Hybrid