PJAOL

AI engineer // CTO // Founder

Patrick
O'Leary

I build search, data and AI systems that hold up in production. Founder of The VGER Group. Irish by birth, DC by choice, open source by habit.

Pulsar map Each line is a project. Point at one.
2007
Built geographic search for Lucene at AOL
~10 yrs
AOL, large-scale web and search
2023
Founded The VGER Group
16
Years pushing to GitHub as pjaol
01

Payloads launched

Things I built because I wanted them to exist, then shipped properly. Real installers, real App Store review, real users. These live right here on this domain.

Payload 01 // Font + Office add-in // v0.1.0

SlideMarks

Harvey balls that behave like text.

If you've ever nudged forty icons into forty table cells the night before a board deck goes out, this one is for you. SlideMarks is one consistent family of 122 presentation symbols packaged as a font, so they size, color and align like any other character in PowerPoint and Word.

  • 122 glyphs in 17 families
  • 1 step installer, Mac and Windows
  • 0 accounts, works offline

Signed and notarized on Mac. Screen reader pronunciation packs for VoiceOver, NVDA and JAWS. SIL OFL font, Apache 2.0 code.

Vendor evaluationQ4 // Draft 3
CapabilityOursTheirsTrend
Data residency
Time to deploy
Audit trail
Vendor lock-in
Total cost

Every symbol above is a character in one font. Click a Harvey ball. It's just text changing.

Data sent to servers 0 bytes, by design
  • No accounts
  • No cloud sync
  • No analytics
  • No third-party SDKs
  • No subscription
  • No server to breach

Payload 02 // iOS app // Live on the App Store

Clio Daye

A period tracker that can't sell your data.

Not "we promise not to." Can't. There is no server, so there is nothing to subpoena, breach or monetize. Your cycle history lives on your phone, behind Face ID, and you can wipe all of it in one tap.

Predictions come from your own history as honest ranges, not a 28-day average that fits nobody. Private by architecture, not by policy, and the privacy checks run in CI on every commit so the claim stays true.

02

Flight path

The short version of how an Irish kid who liked computers ended up building geographic search for one of the biggest sites on the internet, and then a company.

  1. Origin

    Ireland

    Started in university, where it clicked that computing could bridge ideas and opportunity. Never really stopped tinkering after that.

  2. Next hop

    The UK, then the US

    Followed the work across the water, twice. Big systems, bigger teams, and a lot of learning what breaks at scale.

  3. ~ a decade

    AOL

    Large-scale web and search during the years the internet was still figuring itself out. Ranking, relevance, and the endless tug of war between advertising and user satisfaction.

  4. 2007

    LocalLucene

    Built geographic search for Apache Lucene: bounding areas, Cartesian tiers, then a precise distance filter from a center point. AOL contributed it to Apache, it shipped as Spatial Lucene in 2.9, and LocalSolr found its way into Spring, Hibernate, HBase and Compass. NASA JPL folks came knocking about it for ocean and lunar data. Not bad for a filter.

  5. The middle years

    CTO, VP, architect, builder

    Leadership roles across EdTech, HealthTech and startups, including building and managing research and data science orgs. Student evaluation and risk monitoring, conversational patient ML, supply chain analysis, data platforms. The common thread: get the data, make it useful, ship the thing.

  6. 2023

    The VGER Group

    Having worked in small, medium, large and gigantic companies, it was time to build the company I actually wanted to work for. So I did.

  7. Now

    Still shipping

    AI systems for clients, MCP servers, browser extensions, an iOS app and, yes, a font. Same habit as always: if the tool I need doesn't exist, build it well and give it away.

03

The VGER Group

Named for the probe that went out to learn everything it could and came home with it. Same mission, fewer planetary-scale incidents.

Most firms talk.
We build.

A small, senior team doing applied AI, data engineering and production software for health tech, edtech, ad tech and supply chain. Founder-led, federal contractor, Northern Virginia. Senior delivery, no bench padding. The person you meet is the person writing the code.

  • A

    AI Engineering

    LLM systems, RAG pipelines, agentic workflows, model integration. Production, not prototypes.

  • B

    Data Engineering

    Pipelines, warehouses, lakehouses, real-time processing. Built at volume, built to be maintained.

  • C

    Software Development

    iOS to backend. Tested, documented, and handed over clean.

  • D

    Cloud Architecture

    AWS and GCP designed for production load, sane costs, and the team that inherits it.

Reduced our pipeline latency by over 60%.
Bala Health
Senior team, zero hand-holding required.
Three Sixty Corp
They write code, not just strategy decks.
EdTech founder
Exactly the AI expertise we needed to move fast.
Phamily
04

Open signals

Tools I needed, built well, released publicly. Some are fresh, some are older than your intern. Star counts as of November 2026.

05 // Hailing frequencies

Got a problem that needs building, not another deck?

AI that has to survive production, data that has to be trusted, or search that has to be fast. Send it over. I read everything.