Technology Analyst | AI, Data & Business Systems

I turn business problems into working software.

Technology Analyst at Barclays and M.S. Artificial Intelligence student at Johns Hopkins. I connect business operations, data and AI with practical technical execution.

Email meSee the work
0:00 / 0:33

Every shot is a recording of the real software. Screens with demo data say so.

The Keep · draft night on the live board · demo data

less closure-pack prep time with my LLM tool at Barclays
70%src
manager accounts on The Keep, my league app, in weekly use
19src
paying client takes bookings on a platform I built
1src
products I built are live in production today
5src

Selected work

Five products in production, each lit in its own colors.

Every recording is the real software doing something. Where a screen holds someone else's data, it shows demo data and says so.

01Own product · in weekly use

The Keep

Keeper, trade and live-draft system for a 16-team fantasy league. It replaced a spreadsheet that had drifted.

the-keep
Demo dataDraft night: a pick lands on the 16 by 16 board, the clock resets, and Fit zoom shows the whole draft as a color map. A demo draft with invented teams, recorded at real speed.

The league's managers sign in with a personal link, declare keepers, trade players and picks, and ran the 2026 draft on a live board. The league rules live in a tested TypeScript domain layer. The invariants live in Postgres.

manager accounts across 16 teams, in weekly use source
19
SQL migrations, each a reviewed change source
29
test files over a framework-free rules engine source
31

Next.js · TypeScript · Supabase Postgres · Row Level Security · Supabase Realtime · Vercel Cron · Claude API · Vitest

02Client · in production

C2J Auto Spa

Online booking and a back office for a mobile detailing business that used to book by phone and text.

c2jautospa.com
Demo dataThe real booking flow, run locally with the default catalog. The taken 3:00 PM slot is demo data.

Built a booking platform and management console to replace phone and text scheduling. Customers build a cart, pick a vehicle size, choose a real open slot and book. The owner manages bookings, services, customers and daily operations from one console.

paying client takes bookings on it source
1
admin tabs with 14 management panels source
9
merged pull requests source
38

React 19 · Vite · Express on Vercel Functions · Firebase Auth and Firestore · Google Calendar API · Stripe (built, off) · Twilio (built, off) · GitHub Actions

03Portfolio project · live

MarketingOS

Multi-tenant AI software that writes, moderates and publishes a small business's social posts.

marketingos.studio
Demo dataHeld posts wait with a plain-English reason. One approval publishes through the app's single publish path, here to its mock provider. The holds are seeded for a fictional brand.

Built an AI content platform that connects creation, review and publishing in one workflow. MarketingOS generates captions, images, carousels and video, runs each piece through moderation, then publishes it. Approval rules decide whether a post goes out, waits for review, or is blocked.

API route handlers source
74
Row Level Security policy statements across 31 migrations source
44
MCP tools that let an outside model drive an account through the same gates source
6

Next.js 16 · Supabase Postgres and RLS · pg_cron · Claude · fal.ai · Remotion (built, dormant) · Satori · Stripe · Model Context Protocol

04Portfolio project · live

G2 Build Room

Describe a business in one sentence and watch a working app get written, compiled and sandboxed.

g2builds.com
Demo dataType a sentence and the homepage console retunes to your trade. This part runs without a model call, and the console labels its sample data.

A small-business owner types one sentence and immediately gets a deterministic working console for their trade. Claude then streams a real multi-file React app. Nothing reaches the browser until every file compiles on the server, and the app runs in a sandboxed iframe.

measured cost of a full Opus build, 489 lines source
$0.43
AI edits per visitor, counted in a signed cookie source
5
tests, including WCAG contrast pinned to the tokens source
33

Next.js 16 · React 19 · Claude Opus and Sonnet · Server-sent events · Babel compile gate · Sandboxed iframe · Vercel Blob · Resend

05Own product · live

Weekend Golf Warrior

Live scoring, matchups and side games for a golf trip, built for phones with one bar of signal.

Demo dataTwo signed-in phones on one emulator, recorded at real speed. Scores tapped on one phone update the match on the other through live listeners. A simulated weekend.

A captain creates a golf weekend and shares a code. Every player joins from a phone. The app runs the trip: roster, team draft, match play, hole-by-hole scoring from every phone, skins and side games, a chat feed with photos, and a closing ceremony.

simulated phones play a full weekend against the real rules source
19
security-rule tests asserting allows and denies source
100
unit test cases across 37 files source
590

React 19 · Vite · Firebase Firestore, Auth and Storage · Security rules · Firebase Emulator Suite · Vitest · Vercel

Everything else

The shipped ledger, statuses included.

Screens with demo data are labeled where they are shown full size.

Status words mean what they say. Live is in production. Client is someone else's business. Private has no public link, so its screens here are captures. Internal work at Barclays cannot be shown. A public demo is deployed but not counted as a product in production. Every result claimed here has a receipt.

About

Finance experience, technical curiosity, and practical problem solving.

Ryan Gogerty

I studied finance at Lehigh with about 50 credits of mechanical engineering. I now work as a Technology Analyst at Barclays while studying AI at Johns Hopkins. Beyond my professional experience, I have built AI platforms, customer websites and operational tools for businesses and the groups I belong to.

My approach starts with understanding the business problem, gathering requirements and deciding what a useful solution needs to do. Then I build, test and improve it with feedback from the people who use it.

Experience

  1. Aug 2023 to now

    Barclays · Technology Analyst, Wholesale Lending Technology

    • Built an LLM-driven compliance and governance documentation tool on AWS Bedrock with a React front end. It cut closure-pack preparation time by 70%.
    • Organized the stakeholders, coordinated about ten senior leaders over a couple of months, handled the compliance work to get accounts provisioned, and wrote the proof of concept myself.
    • Trained more than 100 employees on Microsoft Copilot and agent workflows in 30-minute sessions, guiding each group through creating a first Copilot agent.
    • Designed and ran a GenAI onboarding tutor as a proof of concept. It did not go to general release, but the prompt approach it proved was picked up by onboarding initiatives across the bank.
    1. InputSource documents
    2. Front endReact
    3. ModelLLM on AWS Bedrock
    4. OutputStandardized closure pack
    5. CheckOne review in proof-of-concept testing
    Schematic. Internal system at a bank, so no screenshots.
  2. Jun 2022 to Aug 2022

    Barclays · Financial Operations Intern

    • Automated financial reporting with Power Automate, saving analysts more than 11 hours a month.
  3. Apr 2020 to May 2022

    GroGo · Founder and CEO

    • Started a grocery delivery business during the COVID shutdown. Served more than 70 families and earned more than $10K in revenue.
    • Led 4 executives and 20 personal shoppers, built the ordering site, and ran operations through to the company's wind-down.

Education

  • Jan 2026 to Dec 2027, in progress

    Johns Hopkins University · M.S. in Artificial Intelligence

    Focus on large language models. Now taking Introduction to Algorithms and Applied Generative AI.

  • May 2023

    Lehigh University · B.S. in Finance, Integrated Business and Engineering Honors

    About 50 credits of Mechanical Engineering and a minor in Economics.

How I work

  1. 01

    Correctness lives in the database.

    A trade in The Keep is accepted inside one locked Postgres transaction. MarketingOS reserves credits in one atomic update. Weekend Golf Warrior rejects writes from stale app builds in its security rules. The UI can be wrong for a second. The data cannot.

  2. 02

    A missing vendor never crashes the product.

    MarketingOS publishes through a provider with a mock default. C2J books without Stripe and skips SMS without Twilio. Weekend Golf Warrior carries on when no course database answers.

  3. 03

    Measure, then decide.

    The Build Room has a measured cost and latency table for each model. NFL Edge poisons future weeks in a test to prove there is no leak, and its docs say the edge is statistically zero.

  4. 04

    Say what is built, what is on, and what is claimed.

    Features that are built but switched off are labeled that way. Every result this site claims has a receipt.

Now

What I am doing this month.

Updated October 2026

  • Taking Introduction to Algorithms and Applied Generative AI at Johns Hopkins.
  • Applying business analysis and AI to workflow improvements at Barclays.
  • Using feedback from real users to improve applications I have built, including C2J Auto Spa and Weekend Golf Warrior.

Contact

Let us talk about what technology can improve.

Email is the fastest way to reach me. If you would rather talk, book 30 minutes.