Mike Vincent Mike Vincent Staff Software Engineer

Mike Vincent

Staff Software Engineer

Los Angeles, California. U.S. Citizen.

Banking, Payments, Infrastructure, AI/ML.

Mike Vincent
Generalist BuilderSoftware EngineerMachine Learning EngineerData EngineerAI EngineerSecurity EngineerPlatform EngineerInfrastructure Engineer

About

Mike Vincent is an American software engineer and founder from Los Angeles who builds and runs payments and AI model platforms for well-known brands. He has worked as an engineering manager and staff engineer at companies including PwC, Capital One, Discover, and Fannie Mae, and writes about Python, programming, and technology leadership in the “Quark’s Outlines” series.

Portfolio

Fraud & Risk Models

Current work. Fraud and risk models for the Discover Global Network, a global payments network. This includes prohibited-transaction detection required under UIGEA and Regulation GG, plus the platform that serves those models in production.

Discover Global Network PwC
Global payments networkFraud modelsRisk modelsModel servingReal-time inferenceFeature pipelinesUIGEA / Reg GGPayments risk

AI & Machine Learning

NLU, entity and intent models behind the Capital One Eno chatbot, running on Capital One’s enterprise AI platform. Transit demand and real time arrival prediction for UMO IQ and NextBus, built in SageMaker with TensorFlow and PySpark.

Capital One Discover UMO IQ NextBus SF Muni
LLMsAI agentsRAGMLOpsKubeflowTensorFlowSageMakerPySpark

Search & Model Inference

Hybrid retrieval for Discover.com and CapitalOne.com over vector embeddings and BM25, scored on offline evaluation sets that measure answer quality. Enterprise intranet search for Lockheed Martin, and hybrid search for Radio Index, an old-time-radio catalog.

Discover Capital One Lockheed Martin Radio Index
Hybrid searchVector searchBM25ANN/KNNAutocompleteOffline evalLuceneSolr

Payments

Fraud scoring on the Discover Global Network, Zelle transfer APIs, Capital One Databolt vaultless tokenization, and contactless iPhone fare payments with Apple for NYC MTA OMNY, all under PCI.

Discover Global Network Zelle Capital One NYC MTA LA Metro Apple
Fraud scoringZelleTokenizationCard dataApple PayPCI DSSFare payments

Infrastructure & Distributed Systems

Three clouds, one set of pipelines. Transit-payment SaaS across five agencies serving millions of riders, with KEDA autoscaling on AWS and Azure.

DC Metro NYC MTA BART LA Metro Boston MBTA
AWSAzureGCPKubernetesTerraformIstioKEDAKafkaRedis

Security & Compliance

PCI, SOC 1 and 2, ISO and SOX audits passed on fare-payment platforms. mTLS with a Smallstep private certificate authority across the Cubic fleet.

Capital One Cubic LA Metro NYC MTA
mTLSPrivate CACrowdStrikeNessusPCISOC 2ISOSOX

Full Stack & Operations

Ships the interface and carries the pager. The Fannie Mae Collateral Underwriter SPA, Poison Control on iPhone and Android built with Cordova, and on-call incident response behind it all.

Fannie Mae Poison Control Canadian Tire USCCB
GoPythonJavaTypeScriptReactCordovaSwiftUISREPrometheusGrafana

Resume

download

Mike Vincent

Los Angeles, CA · U.S. Citizen · mikevincent.dev

Education

University of Texas at Tyler

Master of Science, Industrial Engineering, 2015

The Pennsylvania State University

Bachelor of Science, Linguistics, 2011

Experience

Lead Software Engineer

2024 to Present

PwC · Los Angeles, CA

Founding engineer building AI search and payments infrastructure for Capital One and Discover.

  • Discover Global Network fraud models. fraud and risk models for the payments network, and the platform that serves them in production.
  • Discover.com Search. hybrid search over vector embeddings and BM25, scored on offline evaluation sets that measure answer quality.
  • CapitalOne.com Search. hybrid search with autocomplete and indexing, surfacing results as a customer types.
  • Capital One Eno. NLU, entity, and intent models behind the account chatbot, moved from Kubeflow onto Capital One's enterprise AI platform.
  • Capital One Databolt. vaultless, field-level tokenization in Go and TypeScript. AI models, analytics, and payments run on tokens and never see the real card number.
  • Zelle. onboarding, authentication, and access-control APIs to move money between people in seconds, safely.

Software Engineering Manager, Infrastructure

2021 to 2024

Cubic Transportation Systems · San Diego, CA

Software engineering manager for transit-payment SaaS across five agencies. WMATA, NYC MTA, BART, LA Metro, and MBTA.

  • NYC MTA OMNY. contactless iPhone fare payments with Apple. Kubernetes manifests and Docker containers for the payments app.
  • LA Metro TAP. fare-payment backend as Docker and Kubernetes microservices with vertical, horizontal, and KEDA autoscaling on AWS and Azure.
  • Platform security. mTLS with Istio and a Smallstep private certificate authority. CrowdStrike, Nessus, and Trend Micro across the fleet.
  • Release engineering. Terraform, AWS, and GitHub Actions CI/CD, under PCI and SOC 1/2 compliance. Production code in Python, Java, and JavaScript across AWS, Azure, and vSphere.
  • Ran a team of 18. across the US, India, and Germany supporting transit SaaS for millions of riders. On-call on PagerDuty, monitoring on Prometheus and Grafana.

Lead Senior Software Engineer

2019 to 2021

Cubic Transportation Systems · San Diego, CA

Lead engineer building rider-facing chatbots, prediction, and the cloud platform behind transit fare services.

  • Chicago Metra. rider chatbot built with Twilio and Azure Cognitive Services on AWS Lambda. Riders check transit-card balances by phone, text, or email.
  • Urban Insights. transit-demand machine learning and search in Jupyter and SageMaker, monitored with Prometheus and Grafana.
  • Build and release. automated CI/CD for 20 Java apps with Jenkins, Ansible, Docker, Helm, Vault, Terraform, and Kubernetes on Azure. Database migrations via Liquibase and SQL*Plus.

Senior Software Engineer

2017 to 2019

Cubic Transportation Systems · Washington, DC

First DevOps engineer on Cubic's fare platforms, moving embedded and legacy systems to containers and cloud.

  • NextBus. real-time bus-arrival prediction on AWS ECS and Lambda with TensorFlow, PySpark, CloudFormation, and Node.js.
  • First DevOps engineer. led the transformation for a publicly traded company with Terraform, Kubernetes, and Docker. Embedded fare CI/CD with Jenkins, Conan, and Wind River Linux.
  • Security and compliance. PCI, SOC 2, ISO, and SOX audits. AWS and Azure hardened with Trend DSA, Nessus, and Cisco ASAv, monitored with Prometheus and Grafana.

Senior Software Engineer

2016 to 2017

EPAM Systems · Washington, DC

Senior backend engineer at a digital agency, shipping search, payments, and mobile apps for enterprise clients.

  • Fannie Mae Collateral Underwriter. Angular single-page app on Tomcat with JSON and REST APIs in Python. Appraisers value a home on a mortgage.
  • Lockheed Martin Intranet Search. enterprise search on headless Drupal. Microservices on Docker and Kubernetes with Node.js and Python.
  • Release engineering. Jenkins, TeamCity, and Sonatype Nexus CI/CD on AWS with Chef. PostgreSQL migrations with Liquibase and a zero-downtime multi-tenant switch.

Software Engineer

2015 to 2016

EPAM Systems · Washington, DC

Backend engineer on a full-stack team building mobile apps and e-commerce for national brands.

  • Poison Control. National Capital Poison Center iPhone and Android apps built with Cordova CLI, shipped via App Store, TestFlight, and Google Play.
  • Canadian Tire. Hybris store and the U.S. Conference of Catholic Bishops app on AWS, with RDS and ElastiCache behind Kubernetes and Docker.
  • Infrastructure. Docker and Ansible on VMware ESXi with Ubuntu and CentOS behind Cisco ASA firewalls. Provisioned with Vagrant and Chef.
  • Test automation. Node, Karma, Jasmine, PhantomJS, WebDriver, and Protractor. JMeter and BlazeMeter performance tests. SVN-to-Git migration.

Projects

Io Pictures Corporation - Founding Engineer

Architecture for rights, metadata, payments, storage, and digital distribution systems from zero to production.

The Archive of American Radio, Inc. - Founding Engineer

Architecture for search, RAG retrieval, metadata extraction, AI agent workflows, content processing, and API systems from zero to production.

Skills

Core

Software architectureplatform architecturedistributed systemssystem designsite reliability engineeringSREincident responseon-callbackend servicesAPIs

Cloud

AWSAzureGCPKubernetesDockerTerraformAnsibleCloudFormationPackerHelmIstioKEDASmallstepGitHub ActionsJenkinsTeamCityNexusVaultECSLambdaRDSElastiCacheLinux

Languages

GoPythonJavaTypeScriptJavaScriptSQLPostgreSQLLiquibaseRedisKafka

AI/ML

LLMAI agentsRAGmachine learningMLOpsvector searchhybrid searchBM25ANN/KNNentity extractionintent classificationKubeflowTensorFlowPySparkSageMakerAzure Cognitive Services

Security

mTLSCrowdStrikeNessusTrend MicroCisco ASAPCISOC 2ISOSOX

Search

Search indexingETLoffline evaluationexperiment designLuceneSolrOpenTelemetry