Mike Brannigan

Loti is the only functional deepfake "white hat" and detection system across video, voice, and images on the internet.

Even Google said we couldn't do this. We are.

I contribute as a MLE, Pythonista, and Vector Database Engineer using EMR, Milvus, Jupyter, and all sorts of ML goodies.

This is a very happy place for me, as it sits astride data, do-gooding, and really needs optimization and clever thinking in order to monitor The Entire Internet for malicious behaviour.

Amazing!

My client, Pareto Health, is seeing me perform increasingly sophisticated tasks in the AWS cloud with ML and Data, so I figured I should match up my paper credentials with my actuals. First of Four.

A pivot off of LaVid, Anonifi is launched in the Direct to Consumer space (instead of B2B) to allow people to harness their own PII data and build credible, verifiable, alternative credit profiles, and gain access to better financial products.

Patent Awarded! - For work in the anonymous marketplace, privacy, and applied cryptography space.

US Patent 11,023,942 or view the entire abstract here

Bringing health technology out of the stone age.

One of two Machine Learning Engineers, we acquired and analyzed monstrous EDI data on US Healthcare to inform our underwriting models.

Converted an MS Excel based "manual underwriting" process into a modern mobile and web-based, API-driven CRM adjunct to Salesforce. Enabled a previous bottleneck within the organization to improve its throughput and decision-making 20-fold.

Created re-training pipelines, a data lake, and a reporting system to keep ML manageable, maintainable, and responsive to rapidly changing healthcare trends.

Embarked as a co-founder on a microfinance lending startup, with the intention of ending predatory lending.

The sole in-house technologist, have written and hosted from-scratch numerous microservices, sites, tools, and a complex Ionic hybrid mobile app.

Use custom ML tooling to algorithmically discover "credit opportunities" within a pool of app users who self-collect data on their movements and activities, and submit them anonymously.

The LaVid ecosystem attempts to allow the chronically under-banked to collect alternate lending criteria on themselves, build a cryptographically secure/fraud-resistant credit profile on themselves, and seek financial services on the strength of alternative lending criteria such as utility bills, rent payments, and activity within the community.

On an alarmingly short timeline, I was asked to build a completely PaaS machine learning pipeline to forecast sales, hourly, for the Little Caesar's "Hot n Ready" product as one of a team of three.

I handled all data gathering, aggregation, and prepared a concise and curated ML feed for historical weather, forecast weather, sporting events, POS sales, and other items in order to predict pizza sales per hour.

An enjoyable fire drill, I created the ML support functions, the CI/CD pipelines, and stood up the Azure Timed Functions needed to reliably send predictions to each store via Kafka.

Project was co-sponsored by Microsoft to showcase Azure to their client, Ilitch Holdings, who own Little Caesars and other companies. To support their effort, I also prepared detailed architecture documents, presentation materials, and a detailed DevOps "binder" for management of the system.

I was invited to join a team of 5 developers to convert a legacy/POC project from PHP on PostgreSQL into a more modern RESTful microservice-style API.

I refactored the back-office sales and management portal for a 300-staff Solar installation company.

I incorporated third-party data from Genability to provide accurate energy costs/savings predictions and sales support materials.

I was invited to work on the GlobalCare Analytics engine using Jupyter Notebook.

I ingested millions of patient records in 835/837 format, prepared them for analysis, and then sought to discover predictable insights and risk factors previously unknown to hospital administration.

I used Elasticsearch and Kibana to provide visualizations and business intelligence to administrators.

I maintained the utmost care and consideration for HIPAA-protected patient data, and ensured sanitization of PII and data streams at all times.

I was hired to join the Expensify team, in support of their mobile app product. My goal was to make the "10X" project a success, as we had a few months prior to Expensify's first major Advertising blitz, a Superbowl commercial with artist 2Chainz and a massive promotion of giveaways.

Expensify operated on a "flat" management structure, my first time working within such a thing. Engineers were encouraged to "travel" up and down the software stack, something which delighted me. I was able to work on clustered databases using SQLite and C++, both mobile app codebases, ReactJS/JSX, PHP, and myriad custom in-house tools.

Delivered high-quality peer-reviewed code under a rapid CI/CD system with multiple swim lanes and A/B testing.

Coded custom scripts for in-house DevOps, deployment, monitoring, and colocation tooling.

Hired into the Sony PlayStation Vue group to provide Video On-Demand as a competitor to Netflix and AppleTV.

Worked on tooling for Geolocation using MaxMind data, authentication using J2EE and Oracle, and coded a ReactJS proof of concept to port an aging UI framework to modern web standards.

I was hired to build out UI and tooling in anticipation of online poker legalization in California (Wright, AB 1463)

Joined a team to build online multiplayer poker using a C-based legacy engine from the defunct Full Tilt Poker online casino, which was ruled unlawful.

Built performant real-time gaming using WebSockets RFC 6455 and fallback protocols to support older browsers, devices, or slow internet connections.

I was hired to develop CSOD's video on demand mobile app and bring it to standards compliance.

Supported myriad device types using Apache Cordova and HTML5 Web Standards.

I was hired at a busy Los Angeles movie/entertainment agency to develop mobile apps to support their client advertising, launches, and events.

Sole developer who wrote the NCAA March Madness bracket app for CBS Sports, and the Teenage Mutant Ninja Turtles "Turtles vs Foot" mobile app for Long Beach Comic Con.

I was a member of the Kodak Gallery team, which allowed customers to snap a pic with their mobile device, or even drop a roll of film at their local drugstore, log on online, and order prints or other items.

Worked in an Agile environment using Jira, Greenhopper, Confluence, and Perforce.

I was hired in to build an "App Store" for the BrewMP phone ecosystem, a low-cost phone based on Java.

Wrote all tooling for back-end approval and acceptance of apps; Provided a UI for ratings and reviews, and linked with the CDN for app delivery, changes, or even emergency revocations of apps.

Led a "strike team" of five alongside the BrewMP team of > 50 engineers.