Call for Speakers
in 3 months
DPE Summit 2027
event starts
2 Feb 2027
event ends
3 Feb 2027
location
The Midway San Francisco, California, United States
website
dpe.org/dpe-summit/
DPE Summit 2027 is about the delivery side of the AI-native SDLC, from the agent’s first edit to the release, and about the developers who steer the agents. DPE Summit speakers build, run, or research the systems that verify, review, merge and release agents’ changes at scale.
Since 2022, engineers from Amazon, Airbnb, Google, JPMorgan Chase, LinkedIn, Meta, Microsoft, Netflix, Spotify, Uber and many more organizations have presented their work on builds, tests, CI and developer productivity here.
Come experience the Developer Productivity Engineering Summit with 500 other experienced engineers and engineering leaders who run builds, CI, testing, code review and developer productivity and are paving the way for their organizations. At DPE Summit the conversations are half of the conference, which is why we include multiple breaks and networking opportunities throughout each day.
Hosted by Develocity, DPE Summit is a two-day, two-track conference helping set the stage for engineers and engineering managers working in the AI era.
open, 20 days left
Call for Speakers
Call opens at 12:00 AM
20 Jul 2026
Call closes at 11:59 PM
13 Nov 2026
Call closes in Pacific Standard Time (UTC-08:00) timezone.
Closing time in your timezone () is .
Nobody has the delivery side of the AI-native SDLC figured out yet, including the teams furthest along. That is why they come: to compare notes. We are looking for talks that show what you built, what you measured, what failed, and what you learned.
What talks can be about
- Trusting what agents produce. Regression and specification tests, side effects at agent volume, safety nets for legacy code, formal verification and type systems, checks an agent cannot game, who may change a test, and what an agent may change and reach in CI, from the test settings to the secrets.
- Keeping verification fast and reliable at agent volume. Build and test speed, and CI load. Reducing the cost of building and testing in agent sandboxes. How agents handle flaky tests and failures the change did not cause. Agents that keep the toolchain healthy, finding drift and fixing it.
- What builds and tests tell the agent. Context for agents finding the root cause of a failure, the test suite as the memory of the codebase, and output shaped for agents.
- Review, merge, and release at agent scale. Human and agentic review, which changes merge without a person, merge queues, the cost of reconciling changes with a main branch that moved while they waited, and provenance: proof of how a change was built and what was verified.
- Evaluating the agents. Testing a change to a model, a harness, or the instructions, rules, skills, and tools that steer the agents, before the change reaches every engineer.
- Cost. What a merged change costs in compute, tokens, and developer time, and how to bring it down. The CI bill at agent volume. The right model for each job. The share of compute and token spend that is waste, and budgets and attribution for agent spend across teams.
- From steering agents to supervising swarms. What limits how much one engineer can deliver with agents, from waiting on builds and tests to steering several sessions at once. Who handles the failures agents cannot resolve. What breaks as a few agents become a swarm that a handful of people supervise. Large-scale changes made by agent fleets, such as migrations.
- Running the delivery system like production. SLOs and error budgets for build times and flaky tests, with alerts when they are breached, not only when CI is down. Observability for the delivery system. DORA and delivery metrics, including the measures on the way to production, such as how often the pull request build passes on the first try. Measuring productivity and developer experience when nobody works without AI.
- Securing the toolchain. The toolchain as an attack vector: build tools, plugins, CI, dependencies, and the agents that run with access to them, including the dependencies agents add. Isolation, least privilege, a verified identity for every build and agent, and policy that holds at machine pace. Provenance for every artifact, so that a compromised tool or dependency can be traced to the software running in production.
- Developer experience with agents. Flow, focus, and cognitive load. When developers trust what agents produce, and when they stop. Review fatigue. Which skills developers stop building, and how junior engineers still grow.
Talks on builds, tests, and CI are welcome even when agents are not their subject, if the work is original and still matters. Agents run the same builds and tests that people do.
What we look for
- Work in production at scale, with concrete results.
- Research with data on what works and what does not.
- A design that failed, and what replaced it.
- Deep thinking on foundations.
Our featured past talks show the standard. First-time speakers are welcome, and so are talks based on published research.
How we select
The DPE Summit talk-selection committee reads every proposal and judges it by its depth, its originality, and what others can learn from it. Please use the “note(s) to the committee” field to include any additional information or context you would like the committee to know.
Formats
“Regular” talks are 30 minutes, with time included for questions, lightning talks are 15 minutes, and Keynote talks are by invitation.
As a default, sessions are recorded and later published on the DPE Summit website and used for broader marketing material. If your organization prohibits you from your talk being recorded and/or published, please let the event organizers know.
Every presentation is subject to our DPE Summit Terms & Conditions of Attendance and Participation, Privacy Policy, and Code of Conduct.
The call for papers (CFP) closes on Friday, November 13th. Speakers can expect to hear from us by the end of November.
Speakers will receive:
- Free DPE Summit ticket for themselves
- Significant discount on (1) guest ticket
- Free accommodation at Develocity preferred hotel in San Francisco
- Invitation to Speakers Dinner
accommodation
expenses covered
event fee
free for speakers
If you haven't logged in before, you'll be able to register.
Using social networks to login is faster and simpler, but if you prefer username/password account - use Classic Login.