James Broadhead
@jamesbroadhead
Followers
1K
Following
22K
Media
385
Statuses
15K
Postgres Leader @databricks prev: @neondatabase @mongodb @tweetdeck @twitter. All views my own.
Dublin
Joined June 2009
RT @parys_dariusz: First time working with @neondatabase. I really like the approach of being able to branch.
0
1
0
RT @prmshra: Omg i finally understand why founding engineer and i speak opposite languages . He says “so what are we solving for this machi….
0
77
0
Incredible tech, amazing team, seamless integration - enjoy!.
Introducing Lakebase, the first fully-managed Postgres database integrated with the lakehouse and built for AI. Lakebase combines the familiarity of Postgres with the scalability of the lakehouse and the agility of @neondatabase’s database branching technology. Hundreds of
0
0
5
1
3
35
RT @AndreLandgraf94: Need a Postgres DB for a quick demo, experiment, or random thingy?. Free Postgres - no signup, no credit card, nothing….
neon.new
Instantly hosted Postgres. No config, no account. Perfect for OSS platforms, quick tests, and AI agents.
0
18
0
RT @Codydearkland: I'm so excited to see this hit, its legit every bit as magical as it sounds. I spun up a project over the weekend with….
0
3
0
here's our final writeup about these incidents. Lots of lessons here, especially if you're quickly allocating IPs in EKS. We've filed a PR against AWS CNI. Writeup here --
neon.com
This post covers what we learned about AWS CNI through our post-mortem and root-cause investigation into our outages on 2025-05-16 and 2025-05-19.
0
0
0
Leading Neon Engineering has been a blast! Reflecting on what got us here as we transition into Databricks. Clear Priorities & Customer Focus .Ship to Learn over Perfect First Time .Find Fast, Fix Fast .Finish Things - Throughput over Parallelism.Vertical Slices on Prod
0
0
14
Really great list.
things I believe:. - shipping fast beats the best strategy. - speed is a superpower. - create a bias toward shipping. - small teams ship faster. - ai-native teams will move 10x faster than those not willing to change. - landings > launches (i.e. product adoption >.
0
0
2
We built an app . that builds . Fully OSS AI App Builder. Enjoy!.
app.build
App.build is an open-source AI agent that generates and deploys full-stack apps with Postgres. Use it as a reference to build your own agent.
We just open-sourced a reference implementation for building codegen products on top of Neon: Local-first, built for developers. Inspired by @v0, @Replit, @create_xyz and @samedotdev
0
0
15
Here is our high-level postmortem [1] and a detailed description of the events which triggered the outage [2]. We are continuing to work on a more detailed root-cause analysis, as well as additional detail to add to [2]. [1] [2]
neon.com
For further details, read the top-level Post-Mortem. Summary The Neon Control Plane service is backed by a Postgres database. A scheduled job in the Control plane, Activity Monitor, is responsible...
1
0
4
Mahmoud is the best! Excited to see what he'll do next!.
Personal news: after 2.5 epic years @neondatabase, I’ve decided that it’s time for something new 🫡. I worked alongside an incredible team on a wide range of things - content, shipping demos & driving growth. My DMs are open + here are my highlights ↓.
1
0
13
RT @shreyas: Most people assume that doing good work repeatedly develops good taste. But the causation mostly runs the other way. People w….
0
48
0
RT @bradmenezes: 3/5 Clark is the first AI @databricks Apps Engineer – connecting to Databricks SQL, Mosaic models and @neondatabase, while….
0
3
0
RT @LewisCTech: Watched the original "Functional Core / Imperative Shell" talk from 2012, by @garybernhardt - I'd long known of the idea b….
0
3
0
Over the last year, the team has traced a reliability journey that I am proud of - until this week. I have heard from many of you about disrupted products or engineering teams. We can and will do better. You will hear from us shortly with our post-mortem.
4
0
34
In today’s outage, we continued to see IP allocation failures, even though we reverted the triggering config change and our subnets had available space. We will continue to investigate, and are working on a reproduction case of today’s outage. We are in active contact with AWS.
1
0
12
We took notes of Friday’s incident, including all of the out-of-normal config settings that we had applied in the journey to fixing the incident. We filed action items to revert these back to baseline configuration. Sadly, in reverting one of these settings, we triggered today’s.
1
0
14
Without IPs available, we could not start new computes. Even after we had reduced the number of running computes, IP allocation continued to fail for a protracted period - finally resolving once we hard-restarted the aws-node daemonset.
1
0
14