Easy Day Development AI/ML & full-stack engineering

Hard problems. Easy day.

I’ve spent 16 years building software for defense and government, from Marine Corps personnel systems and a DoD cloud migration to deep-learning research. Through Easy Day Development I work with contractors and other teams by the project or part-time.

Years building production software
16
Java EE applications architected, secured & sustained
56
DoD Impact Level earned in AWS, with an ATO
IL4
Artificial Intelligence & Machine Learning, 2026
M.S.
Certified Information Systems Security Professional
CISSP

01Services

Where I can help

I take on work in four areas, and on every project I’m the person accountable for the result. Bring me in to prototype an idea, rescue a stalled system, or carry a platform all the way to production.

  1. 01

    AI & machine learning

    I build deep-learning prototypes that show whether an idea works before you commit a budget to it, such as generative models for forecasting and computer-vision pipelines for images and documents. If AI isn’t the right tool for the problem, I’ll tell you.

    PyTorch · Diffusion models · GANs · CNNs · OpenCV · Tesseract · Time-series

  2. 02

    Full-stack engineering

    I build production web platforms end to end, with Java and Spring Boot or Python services, Angular front ends, and real-time streams over WebSockets. The code is clean and tested so your team can own it after I leave.

    Java EE · Spring Boot · Python · TypeScript · Angular · REST · WebSockets/STOMP · Protobuf

  3. 03

    Cloud, data & DevOps

    I design serverless and containerized systems on AWS, provision them with Terraform, ship them through CI/CD pipelines, and handle the data migrations that come with the move.

    AWS Lambda · DynamoDB · SQS · S3 · Terraform · Docker · Kubernetes/OpenShift · PostgreSQL · Elasticsearch

  4. 04

    Security & compliance

    I’m CISSP certified and have taken systems through an Authority to Operate. I design security in from the first commit and set up a repeatable process to find, fix, and track vulnerabilities so your authorization holds. That covers DoD STIGs, Impact Level 4 data with PII and PHI, and the documentation reviewers ask for.

    CISSP · DoD STIGs · Vulnerability management · IL4 · Authority to Operate · Secure SDLC

02Lab

What I’m researching now

In recent defense R&D I built a proof-of-concept that applies diffusion models to flight-trajectory prediction. Given where an aircraft has been, the model samples many plausible futures, so you see the spread of outcomes instead of one overconfident line.

  • Observed track
  • Sampled futures
  • Average of the samples
Fig. 2 Reverse diffusion, conditioned on an observed track. Each gold line is one sampled future. If you average them, as a single-answer model would, the path goes straight through the storm. This is an illustrative simulation running in your browser, not a production model.

i.

How it works

During training the model learns to remove a little noise at a time. To sample, it runs that process in reverse, starting from static and ending with a coherent trajectory.

ii.

Why one answer isn’t enough

When an aircraft could go left or right around the weather, averaging the two predictions gives a path straight through it. A generative model keeps both options and how likely each one is.

iii.

Where else it applies

I’m also researching GANs and diffusion models that forecast unplanned work, so schedules can include enough buffer to absorb surprises instead of breaking.

03Selected work

Some of what I’ve built

These are highlights from sixteen years of defense R&D, federal IT, and enterprise engineering, described at the level I can share publicly.

2017 – 2020 · U.S. Marine Corps, Manpower & Reserve Affairs

Moving 56 Marine Corps applications to AWS

As technical lead I owned the architecture, security, deployment, and sustainment of 56 Java EE web applications supporting DoD and congressional mandates. I planned and executed the move from an on-premises datacenter to AWS at Impact Level 4, PII and PHI included, and earned the Authority to Operate under DoD STIGs.

I also set enterprise software standards and DevOps practices, built a repeatable process for finding and fixing vulnerabilities to keep the ATO, led a team of government, military, and contractor staff, and briefed leadership on roadmaps and costs.

  • Java EE
  • AWS
  • DoD STIGs
  • Vulnerability management
  • Microservices
  • DevOps
  • Team leadership
  1. 2021 – now · Two Six Technologies

    Generative forecasting

    I built a proof-of-concept that applies diffusion models to flight-trajectory prediction, a new generative approach to spatiotemporal forecasting. My ongoing research uses GANs and diffusion models to forecast ad-hoc task arrivals and size the schedule buffers that protect planned work.

    • Deep learning
    • Diffusion
    • GANs
    • Time-series
  2. 2021 – now · Two Six Technologies

    Computer vision for document analysis

    I built an R&D prototype for object detection and image classification using convolutional neural networks, with OpenCV and Tesseract preprocessing, to test whether document-analysis workflows were feasible.

    • CNNs
    • OpenCV
    • Tesseract
    • Supervised learning
  3. 2021 – now · Two Six Technologies

    Foundation-model solution architecture

    As solution lead on a government foundation-model (LLM and GenAI) proposal, I reviewed current research and turned input from an interdisciplinary team into one architecture and integration strategy.

    • LLMs
    • Generative AI
    • Architecture
    • Leadership
  4. 2021 – now · Two Six Technologies

    Serverless cross-domain messaging

    I architected a scalable, resilient cross-domain messaging system on AWS using Lambda, DynamoDB, SQS, S3, and REST APIs.

    • AWS Lambda
    • DynamoDB
    • SQS
    • S3
    • REST
  5. 2021 – now · Two Six Technologies

    Real-time visualization framework

    I built an event-driven framework that streams live data to the browser, using Angular 14 and Spring Boot over WebSockets and STOMP, with Apache Ignite and Protobuf underneath.

    • Angular
    • Spring Boot
    • WebSockets
    • Apache Ignite
    • Protobuf
  6. 2021 – now · Two Six Technologies

    Legacy Ruby to serverless Python

    I rebuilt a legacy Ruby application as serverless Python on AWS, with Terraform for infrastructure and CodeBuild and CodePipeline for CI/CD, and moved its data from Cassandra to PostgreSQL with Elasticsearch indexing.

    • Python
    • Terraform
    • CodePipeline
    • PostgreSQL
    • Elasticsearch
  7. 2020 – 2021 · Jacobs

    Monolith to microservices

    I broke a monolithic system into containerized microservices (Docker, Spring Boot, Angular 10, Logstash) and deployed them to Kubernetes on OpenShift, running on AWS with RDS PostgreSQL and Elasticsearch.

    • Docker
    • Kubernetes
    • OpenShift
    • Spring Boot
    • AWS
  8. 2012 – 2017 · U.S. Marine Corps

    Systems Marines and their families rely on

    I architected, built, and deployed Java EE applications for Manpower & Reserve Affairs, including the Exceptional Family Member Program, the Wounded, Ill & Injured Tracking System, and Marine Profile, which gives role-based access to more than 200 data points on every Marine.

    • Java EE
    • Angular
    • REST
    • Role-based access

04Career

What I’ve worked with, and when

My work has moved from SQL and data warehouses to enterprise Java, containers, serverless cloud, and deep learning. Select a role to see which tools it involved.

The interactive chart needs JavaScript. The same data is in the table below.

Fig. 3 Where each tool shows up in my work, by year. Each bar covers a role whose résumé entry names the tool.
View as table
Tools and the roles where they were used
ToolDomainWhere it appears
Diffusion models & GANsAI & MLTwo Six Technologies, 2021–now
CNNs, OpenCV & TesseractAI & MLTwo Six Technologies, 2021–now
LLM / GenAI architectureAI & MLTwo Six Technologies, 2021–now
Java EESoftwareU.S. Marine Corps, 2012–2020
AngularSoftwareU.S. Marine Corps, 2012–2017; Jacobs, 2020–2021; Two Six, 2021–now
Spring BootSoftwareJacobs, 2020–2021; Two Six, 2021–now
PythonSoftwareTwo Six Technologies, 2021–now
REST & microservicesSoftwareU.S. Marine Corps, 2012–2020; Jacobs, 2020–2021; Two Six, 2021–now
WebSockets, STOMP & ProtobufSoftwareTwo Six Technologies, 2021–now
C# / .NETSoftwareSAIC, 2010–2012
AWSCloud & DevOpsU.S. Marine Corps, 2017–2020; Jacobs, 2020–2021; Two Six, 2021–now
Lambda, SQS & DynamoDBCloud & DevOpsTwo Six Technologies, 2021–now
Docker & KubernetesCloud & DevOpsJacobs, 2020–2021
Terraform & CI/CDCloud & DevOpsTwo Six Technologies, 2021–now
DoD STIGs & ATOCloud & DevOpsU.S. Marine Corps, 2017–2020
SQLDataSAIC, 2010–2012; Tiburon, 2017
PostgreSQLDataTiburon, 2017; Jacobs, 2020–2021; Two Six, 2021–now
ElasticsearchDataJacobs, 2020–2021; Two Six, 2021–now
Cassandra & Apache IgniteDataTwo Six Technologies, 2021–now
BI & data warehousingDataSAIC, 2010–2012
  • Degree · 2026

    M.S., Artificial Intelligence & Machine Learning

    Colorado State University Global

  • Degree

    B.S., Computer Science, minor in Mathematics

    Colorado State University

  • Certification

    CISSP

    Certified Information Systems Security Professional

05About

You’d work with me directly

I’m an engineer first. I’ve led interdisciplinary teams of government, military, and contractor staff and briefed leadership on roadmaps and cost, and I still write the code.

I started Easy Day Development so defense and government contractors, and commercial teams that want the same rigor, can bring in a senior engineer who already knows ATOs, STIGs, and Impact Levels without making a full-time hire or paying for an agency. You deal with me, not an account manager, and I stay with the work until it’s running well after launch.

The systems I’m proudest of let people answer their own questions. Command Profile let the fleet compare training and skill sets across battalions on their own, and at SAIC the data cubes I recommended let business users pull enterprise-level numbers themselves. I still help with the hard data questions, but the routine ones shouldn’t have to wait on me.

Read the full résumé

Fact sheet

Company
Easy Day Development LLC
Based in
Denver, Colorado
Practice
AI/ML · Full-stack · Cloud · Security
Experience
16 years, since 2010
Education
M.S. AI/ML · B.S. Computer Science
Certification
CISSP
Engagements
Project · Part-time · Advisory
Replies
Within 24 hours

06Working together

How we’d work together

You can bring me in for a single project, for a set number of hours each week, or as an advisor. Each engagement starts with a conversation, followed by a written plan.

Project

A defined project

We agree on the scope and deliverables up front, whether that’s a prototype, a migration, or a new service. This works best when you know what you need.

Part-time

Ongoing part-time help

I work a set number of hours each week alongside your team, which keeps things moving without another full-time hire.

Advisory

Advice before you commit

I review architectures, check whether an AI approach is feasible, or assess security readiness before you commit a budget.

  1. Discovery call

    We start with a free 30-minute call about your goals, constraints, and timeline. There’s no obligation.

  2. Proposal

    I send a written proposal with the scope, timeline, and price for your project.

  3. Build

    I do the work and keep you updated directly, and you see working software along the way instead of status slides.

Not every problem needs AI. If simpler software is the better answer, I’ll tell you before you’ve paid for the experiment.

07Contact

Tell me what you’re working on

You can book a free 30-minute consultation or just send a note. I reply within 24 hours.

Based in
Denver, Colorado
Working with
Teams across the U.S.
What do you need?
Budget optional
Timeline optional

Your details go to my inbox and nowhere else.