Predictive SplitWinner
Sports prediction for sportsbook trading desks — calibrated, conformal, and anchored to a public tamper-proof ledger before every kickoff. Records get faked in this category; this one can’t be, not even by us.
I get the call when it’s big, late, and broken.
When everyone else is stuck, and shipping looks impossible. A Fortune 10 trading floor. A DoD communications platform. A state records system. A retail program that attracted media attention years before consumer access. The assignment was consistent: enter high-stakes environments, resolve complex problems, deliver a final product, and ensure the team could continue development and support independently after I left. My process in each engagement followed a clear sequence: first, I assessed and addressed the immediate technical challenges impeding progress. Next, I created detailed project documentation and implemented structured knowledge transfer sessions, which included technical walkthroughs and hands-on training, to ensure all stakeholders could understand and maintain the delivered solution. Finally, I established feedback mechanisms to monitor the transition and provided guidelines for best practices adoption. This systematic approach consistently enabled teams to adopt best practices and function autonomously. Three industry firsts resulted—banking, defense, and automotive retail. I never left a contract unfinished and was routinely offered permanent roles, but I always progressed to the next challenge.
Now — SplitWinner
Current work
The only sports prediction feed with a public audit trail.
I build and operate SplitWinner (formerly OffensiveEdge), a platform where each prediction, before any event begins, is hashed and recorded on the Bitcoin blockchain. This process creates a tamper-proof, append-only ledger that anyone can access and independently verify using a data file and standard Python tools. I developed every part of the platform myself, including the data systems, predictive modeling, API, trading desk interface, and verification tools. The distinguishing feature is that users do not have to trust the operator, as all prediction records are transparently available for public audit at any time.
- Verification
- A public append-only audit trail with Bitcoin-anchored timestamps — one pure-stdlib verifier checks the whole chain, no account, no trust required.
- Modeling
- Calibrated probabilities with conformal sets, validated on held-out data and proven in a five-week public live alpha — skips disclosed, nothing deleted.
- MLOps
- Daily training-to-serving pipelines behind a signed, versioned API — parallel model deployment, automated versioning, sub-second inference.
- Product
- Full-stack architecture across Next.js, React, React Native, TypeScript and Python, unifying every product surface on one inference backbone.
The record
Selected work
Every one of these was a contract engagement by design: principal consulting delivered through Robert Half. I was brought in for a specific problem, completed the contract in every case, and was offered a permanent role at the end of nearly all of them. I took the next contract instead, as the permanent offers were worth less than the contract already paid, and the next problem was the more interesting one.
If you want to engage, the process is simple: get in touch with a description of your challenge. We'll book a quick initial call to clarify scope, timelines, and fit.
RumbleOn Integrated Auction Solutions General Motors L3Harris Technologies Brierley + Partners T. Rowe Price Kofile Technologies JPMorgan Chase & Co. Poly Lingo / MedLingo
Shipped the first way to buy a car outright on a major manufacturer’s own website
First to market in production, 2017
General Motors · Shop Click Drive · 2017 Chevrolet, GMC, Buick and Cadillac. Franchise law means the manufacturer can never hold the transaction, so the entire purchase brokers through a third party the customer never sees — one seamless flow built around a legal boundary.Announced years earlier and still not live when I arrived. The program had a 100-person manual QA organization; I wrote the automated suite that gated the release on my own — Selenium, Mocha, Chai, Sinon, SuperTest, Postman — and took Google Page Speed up by 90%.Architected the first Node.js application ever put into service in the Department of Defense
First in DoD
L3Harris Technologies · Communication Systems · 2016 Node.js and React into defense service, on a stack that had never cleared it. Moved 40 remote engineers off strictly-typed C# · test coverage 0 → 80% · React localization framework adopted enterprise-widePioneered the first single-page application in banking — chase.com and the Chase mobile app
First in banking top 6 of a 300+ division
JPMorgan Chase & Co. · CMH1 flagship team · 2014 One of six engineers on the flagship team at a Fortune 10 bank. Built on Ember, when a bank shipping a single-page app at that scale was still unproven · jQuery fully removed · codebase down 30% · ten minutes off every buildChief engineer on a national live-auction platform — sub-two-second feed, bids exact to thirteen decimal places
< 2s 13 decimals
Integrated Auction Solutions In a live auction, rounding error decides who wins — thirteen places of precision removed it for every bidder in the country.Built the consumer website, the mobile app, and the on-site inventory cataloging apps outright — React and React Native on one schema.Principal engineer for the largest motorcycle retailer in the country — web, mobile, and a serverless rebuild
−75% cloud $40K → $10K / mo
RumbleOn Created Carvis, the direct vehicle-purchasing mobile platform, and led website and mobile development end to end — AppSync and GraphQL on React Native, lead-to-sale conversion up 22%.RumbleOn ran on the auction platform I was chief engineer on at Integrated Auction Solutions, and brought me in as principal engineer to own their side of it.Shipped the localization layer behind a global asset manager’s client site — the languages its brokers use with clients
15+ locales 500+ a11y fixes
T. Rowe Price Granite i18n across every supported market, and the ADA remediation pilot that found and resolved 500+ accessibility failures on the same platform.Brought the State of Texas digital records system into ADA and Section 508 compliance.
Statewide task completion +34%
Kofile Technologies Rebuilt the experience from business-centric to human-centric, and taught the product managers to treat accessibility as a requirement rather than a remediation.Half-owner of an award-winning EMS interpretation company, through to its acquisition
50% partner
Poly Lingo / MedLingo → ESO Solutions Texas Fire Chief’s Lone Star Achievement Award · EMS World Top Innovation Award · shipped the iOS product to the App Store.Developer tools shipped in the open, growing entirely by word of mouth
76K+ installs and climbing
10 VS Code extensions · 2 Rust binaries 13,700 in the first week and it never slowed down — no marketing, no launch budget. Currently porting the suite to Zed and republishing the cores as Rust crates. letools.devcrates.io
Open source
MIT · in public
pixelcoords
Freeze your screen, mark regions with real shapes, and get pixel-exact coordinates machines can use — versioned JSON, labeled crops, click code, verification with exit codes, self-healing relocation.
Rust MITmacOS · Windows · Linuxcrates.io
pixelactions
Verified desktop interactions driven from a pixelcoords session — click, type, chord, drag, scroll at human-marked coordinates, then confirm they landed. Chained CLI, flow files, or a line protocol any language can drive.
Rust MITmacOSLine protocol
The *-le suite
Ten zero-hassle VS Code utilities, past 76,000 installs. Secrets detection that never leaves your machine, env sync with visual diff and conflict resolution, regex with live previews and ReDoS scoring, and seven more. Being ported to Zed, with the cores republished as Rust crates.
TypeScript 10 extensions76K+ installsZed & Rust in progress
Credentials & stack
What I actually run
- Post Graduate Program, Artificial Intelligence & Machine Learning UT Austin Business applications of AI/ML — the formal grounding under the platform work that followed.
- US patent, multilingual emergency medical communication US 20100223050 A1 Held as a partner in Poly Lingo; filed by founder Ken Kelly.
- MLOps
- MLflow
- scikit-learn
- TensorFlow
- PyTorch
- ONNX
- Probability calibration
- Conformal prediction
- AutoML
- Agentic systems
- LangChain
- LangGraph
- Langfuse
- MCP
- RAG
- pgvector
- TypeScript
- Python
- Rust
- React
- React Native
- Next.js
- Svelte
- Node.js
- Bun
- GraphQL
- PostgreSQL
- Redis
- Kafka
- n8n
- Docker
- Kubernetes
- Terraform
- AWS Lambda
- AWS SageMaker
- EC2
- Serverless
- Microservices
- CI/CD
- Tailwind CSS
The obvious questions
Answered flat
- “Three industry firsts. Says who?”
- The public record covers what shipped and when—launch dates, press, filings—those are easy to check. My role inside those rooms isn’t public, and that’s how the industry works. If you want a resume line you can independently verify, look to the open source: it’s all there, every commit, every decision, every mistake, and every fix in daylight.
- “So what can I actually check?”
- The open source, completely. Ten repositories with full history, CI you can watch run, releases with provenance you can verify, packages on four registries. The install figures come from the registries' APIs, not from me. Start anywhere.
- “Top 3% of Cursor users. Do you write the code?”
- I write the constraints. Every repository carries an agent guide encoding house standards, and generated code either matches or it doesn’t land. The gates don’t care who typed it—a failing scan fails the build either way. The standards are enforced by build gates and checked into every repo. Not assertion—receipts.
- “Has any of this ever gone wrong?”
- Yes, and it’s on the permanent record. Six weeks into the SplitWinner alpha a schema migration rewrote an anchor that was already published. I broke my own append-only rule. That day’s timestamp can never be made to bind again, and the ledger says so, forever.
That work is authentic and now inaccessible. While it is possible to verify that a bank launched a single-page application in 2014—an uncommon achievement at that time—there is no opportunity to review the underlying code… The more significant the environment, the less of the work can ever be publicly shared.
The bar
Claims match runs
These aren’t aspirations. Everything is written into the agent and contributor guides in my repositories and enforced by a check that fails the build.
Zero Guesswork
Relentless Testing Everything I build is tested to break, not just to pass. I take an adversarial approach—writing failing tests first, enforcing checklists in CI, and proving correctness before anything ships. No “it should work”—if it’s not tested, it doesn’t count.Only What Survives
Pragmatic, Not Dogmatic I’ve seen every methodology sold in the last decade and tried most in production. The right tool is the one that clears the bar for reliability, not the one with the loudest supporters. I cut dependencies, pin versions, and treat every external as a liability until proven otherwise.Tried It All
10,000+ Hours Agentic Expertise Years spent building, breaking, and fixing autonomous systems—enough to see every shortcut, failure mode, and edge case up close. Every repo carries standards and agent guides, checked in and enforced by CI. The receipts are public: open source, reproducible builds, and a changelog that logs the bad as well as the good.From Fortune 10 to Solo
Shipped What Works From legacy rewrites and greenfield launches to million-line monoliths and solo open source projects—I’ve run the full gauntlet. The only standard that matters: does it survive first contact with reality?
Elsewhere
All of it is me