←
David Poll2,728 posts
David Poll avatar

David Poll

@depoll

Leading engineering for PRs and Copilot Code Review @ GitHub. Previously Firebase, Parse, Google Cloud, Microsoft. I like building things developers love.

◷ Joined January 2009▧ 52 Photos and videos
Camilla Moraes avatar
⟳ David Poll retweeted
Camilla Moraes@moraes_c_·Nov 10

A few GitHub dashboard improvements just shipped based on user feedback: - Pin the Feed or Home tab as your default view - 6 Issues & PRs shown by default (up from 3) - More compact lists with single-line titles - Agent sessions section is hidden if no sessions exist Learn more: github.blog/changelog/2026-1…

Mario Rodriguez avatar
⟳ David Poll retweeted
Mario Rodriguez@mariorod1·Nov 10

We’re building the new PR page in the open, with all of you. It’s opt-in for now because there’s still plenty to do. Keep the feedback coming both ways ❤️ 😠

the new github PR overview is a welcome improvement

Post attachment
David Poll avatar
David Poll@depoll·Nov 10

So glad folks are enjoying the work the team's been doing. We're really excited to keep working and improving based on all of your feedback!

the new github PR overview is a welcome improvement

Post attachment
David Poll avatar
David Poll@depoll·Nov 5

Confession: I’ve scrolled past dozens of comments on a PR just to find out why I couldn’t merge it. More times than I’d like to admit. Agents have made this a lot worse. They push commits and leave review comments of their own, and all of it lands in the timeline. PR timelines have gotten dramatically noisier, and the few things you actually need to know get buried. My team is rethinking the foundations of the PR experience around that problem, and today the first piece went into public preview. The new Overview page puts merge readiness and your next step right at the top. A summary catches you up on review activity, and you can filter the conversation down to unresolved threads. It’s opt-in and rolling out through the weekend. This is a much-needed start, and there’s a lot more coming as we modernize PRs for the way software gets built now. If you try it, we’d love to hear what still sends you back to the classic view. github.blog/changelog/2026-1…

0:24
Vlad F avatar
⟳ David Poll retweeted
Vlad F@Vlad_GitHub·Oct 9

I got an apology note from the team. We accidentally published some benchmark results but at least we are number #1 :) More seriously open benchmarks helps us make sure we deliver great value so thanks to @computesdk

AI Gateway Latency - Kimi 🥇@github 🥈@neondatabase 🥉@ngrokHQ

0:17
Sameen Karim avatar
⟳ David Poll retweeted
Sameen Karim@sameenkarim·Oct 8

Approvals are now retained for Stacked PRs after stack rebases ✅ Whether you use the "Rebase stack" button or partially merge a stack, as long as the diff doesn't change, PR approvals will be kept even if you have "dismiss stale approvals" turned on

1:02
Ankit Jain avatar
⟳ David Poll retweeted
Ankit Jain@ankitxg·Oct 5

Want to discuss code review with @depoll, Senior Director of Engineering at GitHub, where he leads the team behind pull requests and Copilot code review? Join us for our next The Hangar session: docs.google.com/forms/d/e/1F… We'll discuss: - Why code review was never really about correctness, and what its real jobs are when agents write more of the code - How upfront planning—PRDs, design docs, architecture reviews—is collapsing into the PR itself - Where reviewers should direct their attention when reading every line no longer scales - Turning review decisions into rules and context that become institutional memory for both people and agents - Preparing for a world with 20x more PRs, many of which will never be merged

Post attachment
David Poll avatar
David Poll@depoll·Sep 23

Take UI. For a consumer app, taste in the interface might be the product. For a line-of-business app, I'd often happily let the AI build it. The capability is the same both times. What changes is whether a person gets or adds anything by being involved.

David Poll avatar
David Poll@depoll·Sep 23

On the getting-value side, sometimes being in the loop is the point. Reviewing a teammate's change teaches you where the system is heading, and a better model doesn't change that.

David Poll avatar
David Poll@depoll·Sep 23

The sequence I use now: what are we trying to accomplish and why does it matter, where do people add or get value by being involved, and only then what the AI can handle and with how much oversight.

David Poll avatar
David Poll@depoll·Sep 23

Anything you build on today's capabilities can get absorbed by the next model release. What people value in a product is much more durable, so that seems like the safer thing to design around.