Xavier Briggs

Portfolio

Xavier Briggs

A solo engineer who designs, builds, and maintains production systems. Notre Dame CS student, returning Amazon Prime Video intern, and freelance owner-operator.

  • Amazon Prime Video - returning SDE intern
  • Freelance owner-operator - TLC's production stack
  • Forecasting and trading systems - FORTUNA and Aeolus, built solo

Project 01

in production, 3 properties

TLC

Sole engineer behind a manufactured-home finance company's production stack: I designed, built, and run its three web properties, 11-person CRM, and lead funnel. Monthly inbound leads grew about 5x (~20 to ~100).

Three production properties - a borrower site, a 197-dealer portal, and an 11-person CRM - run on TypeScript, React/Next.js, Firebase, and Supabase/Postgres, backed by a Twilio-based voice intake agent.

Inbound leads · per month

~20grew to~100

before the funnelafter

5x growth in monthly inbound leads, funnel built end to end

3 web properties designed, built, and operated solo

Borrower site
~2,500 monthly visitors
Dealer portal
197 dealers on the network
Internal CRM
11 person team, daily use
Voice intake agent
answers, qualifies, and hands borrower calls to the CRM

TypeScript · React · Next.js · Firebase · Supabase/Postgres · Twilio ConversationRelay

Read the case study

Project 02

commercially launched, operating

Aeolus

A 26-station temperature-forecasting API. Statistical calibration cut forecast error 45% (CRPS 2.25 to 1.23), and it runs unattended at 99.8% daily reliability.

EMOS statistical calibration corrects GEFS ensemble output station-by-station, cutting forecast error (CRPS) 45% through iterative recalibration.

CRPS · forecast error, lower is better

2.25raw GEFS ensembledown to1.23after EMOS calibration

45% forecast-skill improvement, verified before live deployment

calibrated forecasts issued
20,000+
automated daily success rate, unattended operation
99.8%+
stations covered by the forecasting API
26

Python · GEFS ensemble data · EMOS calibration

Read the case study

Project 03

shadow mode, 0 live trades, by design

FORTUNA

An autonomous prediction-market trading platform that forms its own calibrated probabilities and runs every decision through layered GO/NO-GO gates. 3,400+ trade decisions evaluated in shadow mode.

Three independently operating systems - Alexandria, Pythia, FORTUNA - enforce seven regulatory-inspired invariants in code so no system grades its own output.

  1. Alexandriaincorruptible data witness
  2. Pythiabelief formation
  3. FORTUNAjudging and execution

- no system grades its own output

shadow-mode trade intents evaluated
3,400+
regulatory-inspired invariants enforced in code
7
independent systems, none grades its own output
3
calibrated-probability pipeline, every fetch logged and hashed
5-stage

Python · Rust · Postgres · Anthropic API · Kalshi API

Proof from the field

What the work moved.

5xInbound leads

monthly leads, before and after (~20 -> ~100/month)

3Production properties
web properties designed, built, and operated solo
197Dealer network
dealers served by the portal
11CRM team
person internal team using the CRM
LiveVoice agent
production voice intake agent answering inbound borrower calls

About

The path behind the work.

Education

  • University of Notre DameMay 2027
    Notre Dame, IN

    B.S. Computer Science and Engineering, ITREDS Scholar - GPA 3.91

  • University of GalwayJan 2026 – May 2026
    Galway, Ireland

    Study abroad - First Class Honours in all coursework

Experience

  • Amazon Prime Video - Apple Player TeamMay 2026 – Aug 2026
    Software Development Engineer Intern (returning) · Seattle, WA

    Productionizing a tvOS media playback feature toward general availability, owning concurrent multi-player support across Swift, SwiftUI, and AVKit.

  • Amazon Prime Video - Apple Player TeamMay 2025 – Aug 2025
    Amazon Propel Intern · Seattle, WA

    Built a tvOS media playback feature from scratch across 4 phases, 10+ merged reviews with zero regressions, and cut Team Knowledge Agent context usage from 12% to 5%.

  • TLC Manufactured Home Finance Consultants2024 – Present
    Lead Developer (Freelance, Full-Stack) · Remote

    Sole engineer for three production web properties and an 11-person CRM; grew inbound leads ~5x and shipped a Twilio-based voice intake agent.

  • Tweeds - Personal Publishing PlatformJan 2025 – May 2025
    Backend Developer · South Bend, IN

    Built RESTful controllers powering Gemini-driven content analysis and a manuscript-to-marketing generation pipeline.

Leadership & programs

  • ColorStack at Notre Dame · President

    Leads a 25-member chapter for underrepresented CS students, running 5 events a semester.

  • CS4Good · Treasurer (previously President)

    Led 8 community software projects for assistive schools, community centers, and nonprofits.

  • Management Leadership for Tomorrow · Career Prep Fellow

    Selected into an 18-month professional development program for high-achieving diverse talent.

  • AI4All Ignite · Changemaker

    Built an end-to-end deepfake speech detector (81.5% accuracy), presented at the Ignite symposium.