Work

EnChannel builds medical devices, which means every software artifact eventually has to stand up in front of the FDA. When I arrived, builds took about an hour and nothing about them was reproducible.

I re-architected the pipeline and spread the work across a fleet of self-hosted runners, which took builds from roughly 60 minutes down to about 5. Docker made the toolchain deterministic, so a build on a developer’s workstation produced the same result as one in CI. The end goal was byte-level reproducibility, for IEC 62304 verification and FDA submission.

I also implemented SPDX SBOM generation and audit in CI for every release artifact, aligned with FDA premarket cybersecurity guidance and traceable to upstream CVEs.

All of it ran inside FDA-regulated, IEC 62304-aligned lifecycle constraints: chain-of-custody, access control, traceability, and audit-readiness at every stage, with the build fleet held to the same controls.

Two years consulting for a small portfolio of remote clients in ticketing, e-commerce, and competitive intelligence. The work was mostly one shape: multi-region collection infrastructure moving millions of events a day, and keeping capture rates steady while anti-bot vendors shipped updates meant to break exactly that.

When new challenge classes appeared (TLS fingerprinting, JS challenges, behavioral signals) I reverse-engineered them as they landed and rebuilt client collection layers to match, without losing throughput.

I also worked locally with FUTO on tooling, infra hygiene, and applied-security work for their FOSS-aligned consumer software. They get a fuller shoutout on the about page; they’ve earned it.

The rest of the period went to focused study of regulated-device software lifecycles, build reproducibility, and supply-chain provenance, which became the foundation for the EnChannel build-engineering role.

I joined Automatiq as the second platform hire and inherited a collection setup that was a handful of EC2 boxes held together with hope. Over three years I rebuilt it into a distributed platform handling 10M+ collection events a day across Ticketmaster, AXS, and SeatGeek.

I owned the platform’s capture rate, which in this business is the whole product. That meant reverse-engineering Akamai Bot Manager, PerimeterX, and Datadome, and keeping collection alive through vendor updates that took competitors offline. The improvements I shipped were credited with roughly $5M in annual revenue.

The thing I’m proudest of is Transferless Delivery, a resale primitive I co-invented with one other engineer in a matter of days; it became company IP and shipped as a product. Instead of transferring a ticket out of the seller’s account, it delivers the buyer a secure access link to the live mobile ticket, rotating barcode and all, while the barcode stays in the seller’s inventory until it’s scanned at the venue.

That removed the ownership-transfer step that gated competitor business models: near-instant fulfillment with proof-of-delivery logging, thousands of orders, and a new revenue line. It also cut double-sale and transfer-failure risk, handled the cases account-to-account transfer couldn’t (hard tickets, partial barcodes, non-ingested POS listings), and integrated with TM Resale and TM+ pending-sale protection.

Underneath: I led the migration from monolithic EC2 to containerized AWS Fargate, which cut infra cost about 60% and got deploys under 5 minutes. The pipelines were event-driven on SQS, Lambda, and DynamoDB, with circuit breakers, exponential backoff, and DLQ-based retry.

I stood up the observability stack (CloudWatch, Datadog, Logflare) and owned the on-call rotation for the scraping platform. A system that started with no formal incident-response baseline ended at a post-detection MTTR of one to two hours. There’s a longer write-up of the collection work on the projects page.

MINDWISE grew out of Track4: a real-time threat-intelligence platform protecting cardholder exposure across a portfolio of U.S. financial institutions and major payment processors. I co-founded it and ran engineering as CTO.

I hired and led a team of four engineers from initial idea through enterprise rollout, and personally wrote the core ingestion service in Node.js: concurrent scrapers that captured darknet market listings as they were posted and flagged compromised cards within minutes.

The work I’m best known for there is identifying the point of compromise in the Jason’s Deli breach by clustering 170,000+ stolen cards from Joker’s Stash against merchant-location data. The methodology and visualizations were published by Krebs on Security.

We integrated LexisNexis for high-confidence identity resolution, which turned partial cardholder data into actionable intel, and the early detection prevented significant fraud losses for Golden 1 Credit Union.

We were also an operational partner to the U.S. Postal Service Cybercrime Unit, and built vendor-attribution systems out of shipping analytics and geospatial triangulation to support federal investigations. More on the platform itself is on the projects page.

Track4 was the first version of what became MINDWISE. I co-founded it as a threat-intelligence startup, and we landed First National Bank of Nova Scotia as the first client.

The product was automated surveillance of the criminal marketplaces where stolen card data changes hands, turned into reports the bank could act on before compromised cards were exploited. Six months in, it pivoted and grew into MINDWISE.

At Gaine I built core components of the Sanator Provider Registry, California’s mandated healthcare provider registry under SB137, now the statewide industry standard for cross-agency medical data exchange.

The work itself was ETL pipelines and data-quality frameworks over millions of provider records, all under HIPAA and PII controls, with audit trails and governance for secure exchange. It was my first taste of building software where the regulator is a first-class user, a thread that resurfaced a decade later at EnChannel. There’s a longer write-up on the projects page.

I joined Azorian at sixteen, one of a three-person red-team cell running pentests and web-app vulnerability assessments against enterprise infrastructure. My specialty was JavaScript exploitation and auth-bypass classes.

I built custom Python and .NET tooling for reconnaissance, exploitation, and lateral movement. On several engagements we recovered stolen client assets by turning the adversary’s own infrastructure against them.

A summer embedded with a ten-engineer spacecraft team at NASA Ames, working on a satellite imaging payload.

I designed and prototyped a Uniform Light Source for infinite-focus lens calibration on the payload (a mission-critical component) and pitched in across the electrical and software systems wherever an extra pair of hands was useful.