Gabe Ochoa

I build things people use, and the teams that ship them.

I lead Transactional Storage at Datadog, which is nine teams and 66 engineers running the application data platforms that sit under every Datadog product: Postgres, an in-house key-value database, search, replication, and caching. It is the layer nobody thinks about until it stops working.

I came out of Georgia Tech as a mechanical engineer and I wanted to build physical things, so I went to Quirky and spent my days prototyping for a four-person team that reported to the CEO. From there I moved to Wink, where I spent a year taking fifty-odd partner products apart to work out why they didn't do what they promised. Somewhere in the middle of all that I started actually writing code, which was not the plan. Then I was running the infrastructure underneath it at Squarespace, then all of engineering at Numina, and eventually leading the teams that build the thing instead of building it myself. The title changed every couple of years. The work didn't: take it apart, figure out how it actually works, put it back together better.

I'm still happiest where software meets the physical world, where the data is about something that is genuinely happening out there rather than a number that only exists in a database. And I care a great deal about being the kind of manager people do their best work for, which is harder to put a number on than anything else on this page.

You can email me, or find me on LinkedIn and GitHub.

How I work

First 90 days anywhere, the order is always the same: meet the customers, read the code, find the money, and ship something small enough that it actually ships.

Work

  1. Datadog · Senior Engineering Manager, Transactional Storage 2025–present

    I run the org that owns Datadog's application data storage: Postgres, an in-house key-value database, search, replication, and caching. The last stretch has mostly been about cost and consolidation. We cut seven figures a year of infrastructure spend in a single quarter, reset a rewrite of the KV storage engine that had stalled and got it to GA, where it now absorbs 2× load in under 10 minutes at 30% lower cost, and retired two legacy database platforms so the teams had fewer things to keep alive at 3am.

  2. Datadog · Senior Engineering Manager, APM Data Observability 2023–2025

    Led the 14-engineer group behind the APM Data Observability suite, where the job was to build for someone Datadog hadn't really served before: the data engineer. We incubated two products for them. Data Streams Monitoring went from pre-revenue to multi-million-dollar ARR in twelve months, and Data Jobs Monitoring went from private beta to GA with double-digit month-over-month growth. We doubled the team while doing it, which took 180+ interviews and a lot of weeks where hiring and shipping were the same job.

  3. Numina · Head of Engineering 2022–2023

    Owned engineering, product, and R&D for edge-AI computer-vision sensors, the data platform and public API behind them, and the custom hardware itself. We got data accuracy up 20–50% through model retraining and algorithm work, cut shipping time from months to under a week by fixing how code got delivered and tested, and redesigned the pipeline to absorb 3× ingestion growth. Small enough company that I also did inventory planning and handled the contract manufacturers, which I enjoyed more than I expected to.

  4. Squarespace · Engineering Manager → Senior SRE / Team Lead, Infrastructure 2017–2022

    Led five infrastructure teams across compute, traffic, storage, data center, and security. The two I'm proudest of: architecting a performance rewrite of the core edge load balancer, which made every secure connection to every Squarespace site 5× faster, and migrating 3,000 VMs to bare-metal Kubernetes, which took deploys from an hour and a half down to five minutes. Also drove a 3× gain in data-center efficiency, and taught the Incident Commander training and Code Camp.

  5. Earlier · Wink and Quirky 2014–2017

    Site reliability at Wink, where I moved 18 applications off Heroku onto Kubernetes on AWS in three months and learned most of what I know about running things in production. Before that, prototyping at Quirky on the CEO's four-person BlackOps design team, building the prototypes for Quirky Aros and Wink Relay and running vendors in Asia through the last stages of R&D on new smart-home products. That job is the reason I still like hardware.

The full history with dates and detail is in the résumé (PDF).

Projects

Built by Experiment is the nights-and-weekends lab, and it exists so that I keep my hands on 0-to-1 work instead of only talking about it in meetings. Everything below is designed, built, shipped, and supported by me.

Contact

Talk to me about senior engineering leadership, advisory work, and board seats.