Skip to content
Mitchel Carson

U.S. Air Force veteranActive TS/SCIAustin, Texas

Mitchel Carson

Software Engineer · Applied AI & Scientific Computing

I chose to pursue AI while flying executive missions for the U.S. Air Force. Now I build applied AI that makes complex data useful to the people relying on it.

UT Austin M.S. Artificial Intelligence student (4.0 GPA). I built production Java and Spring Boot APIs at USAA, and I research watershed forecasting with deep learning, work motivated by living through Hurricane Helene in Boone.

Schedule a conversationRésumé (PDF)Read the story
Portrait of Mitchel Carson
Executive airlift missions
50+
Including Air Force Two · 30+ countries
Safety-related incidents
Zero
Every mission, 89th Airlift Wing
Less troubleshooting time
30%
USAA · Summer 2025
GPA
4.0
UT Austin M.S. AI · May 2027

01 · Service

U.S. Air Force · 89th Airlift Wing · 2020–2023

Before I wrote code, I flew executive missions aboard Air Force Two.

From 2020 to 2023 I was an Executive Missions Aviator in the 89th Airlift Wing at Joint Base Andrews, the wing whose executive airlift mission supports the Vice President, the First Lady, the Secretaries of State and Defense, and the Chairman of the Joint Chiefs of Staff. I flew aboard Air Force Two, the call sign of the aircraft carrying the Vice President.

It was while serving that I chose to pursue AI. The job taught me what it means for people to rely on your work, and that is the standard I build to.

The job

  • Onboard with senior leaders

    Worked directly onboard with distinguished visitors, including the Vice President aboard Air Force Two and the Secretary of State.

  • Safety and logistics

    Managed passenger safety and logistics with flight crews, security teams, and White House staff.

  • Mission administration

    Handled in-flight meal and beverage service, baggage logistics, crew hotel bookings and ground transportation, visa applications, and crew and passenger subsistence billing.

  • A zero-incident record

    More than 50 executive airlift missions across more than 30 countries, all completed with zero safety-related incidents.

What I carried into engineering

  1. 01

    Preparation is the job.

    A mission was planned long before anyone boarded, down to the hotels, the visas, and the in-flight service. I build research the same way: data manifests, configuration, and leakage checks settled before a model trains.

  2. 02

    Communicate up, down, and across.

    I worked alongside flight crews, security teams, and White House staff. Now it is product owners, engineers, and scientists, and the habit is the same: say what is known, what is not, and what happens next.

  3. 03

    The standard does not move.

    More than 50 missions, zero safety-related incidents. I hold my own work to that bar: evaluation that matches how forecasts are really used, and no result published before the analysis supports it.

I hold an active TS/SCI clearance and welcome conversations with defense, intelligence, and other mission-driven AI teams.

02 · Path

The path so far.

Four chapters with one through-line: work other people depend on, done with careful preparation and a standard that does not move.

  1. 2020 – 2023 · Service

    U.S. Air Force

    Executive Missions Aviator · 89th Airlift Wing · Joint Base Andrews, MD

    50+ executive airlift missions across 30+ countries with zero safety-related incidents, including Air Force Two missions with the Vice President and missions with the Secretary of State. This is where I chose to pursue AI.

    Service record

  2. Graduated December 2025 · Foundation

    Appalachian State University

    B.S. Computer Science · Cum laude · Boone, NC

    A Data Science Certificate and a senior honors thesis on runoff forecasting with deep learning. Experiencing Hurricane Helene in Boone is what motivated HYDRA, my research on watershed dynamics and forecast reliability.

    Thesis-era code

  3. Summer 2025 · Industry

    USAA

    Software Engineering Intern · Global Headquarters, San Antonio, TX

    Java and Spring Boot GraphQL APIs for customer-data workflows used across enterprise channels. API enhancements and JavaScript comparison views cut troubleshooting time by 30%.

    Case study

  4. 2026 – May 2027 · Research

    The University of Texas at Austin

    M.S. Artificial Intelligence · 4.0 GPA · Austin, TX

    Machine learning, deep learning, reinforcement learning, and AI ethics completed. HYDRA continues as ongoing research, with a results manuscript in preparation and a software paper planned.

    Coursework

  5. December 2026 · Upcoming

    AGU26 Annual Meeting

    Scientific workshop facilitator · San Francisco, CA

    Leading an accepted workshop on best practices for AI and agentic workflows in earth science research.

    Research and speaking

03 · Work

Selected work.

Research I lead, production software I shipped, and systems I am building. Each project links to a case study with the problem, the approach, and what is still open.

All case studies

Featured research

HYDRA

Watershed forecasting research with deep learning

Experiencing Hurricane Helene in Boone is why I study watershed dynamics and forecast reliability. HYDRA asks whether a deep-learning model can make NOAA’s NextGen streamflow forecasts more reliable 1 to 18 hours ahead, using only information that would really be available when the forecast is issued.

In progress

Results pending · results manuscript in preparation for Water Resources Research; software paper planned for Environmental Modelling & Software

Research detailsCase studyCode on GitHub

How it works

  1. Generate

    Reforecasts and data

    NextGen reforecast generation software and Google Cloud workflows acquire, validate, and align weather, streamflow, and forecast data with traceable provenance.

  2. Correct

    A learned post-processor

    Designing post-processing for 1–18 hour lead times, comparing LSTM, Transformer, and Mamba-style models on identical inputs and splits.

  3. Evaluate

    By site and lead time

    Leakage-aware temporal splits and hydrologic metrics (RMSE, NSE, KGE), reported by site and lead time.

  • Industry

    Production

    USAA

    Enterprise GraphQL APIs

    30%less troubleshooting time

    Built Java and Spring Boot GraphQL APIs for customer-data workflows used across enterprise channels, plus API enhancements and JavaScript comparison views that integrate name and employment data from multiple sources.

    • Java
    • Spring Boot
    • GraphQL
    • JavaScript
    Case study: USAA

  • Systems

    Active development

    Harmony

    Automated decision system with bounded LLM autonomy

    A financial decision pipeline built as an explicit state graph with declared action paths and tamper-evident logging. LLMs sit in advisory nodes that can flag or veto but cannot execute, and every consequential action requires human approval.

    • Python
    • scikit-learn
    • SQLite
    • React
    • AWS
    Case study: Harmony

  • Client work

    Delivered

    GreenSpace Lawn Care

    Small-business website and digital strategy

    Built and launched the company’s website and advise on its social media strategy, translating business needs into a clear digital experience.

    • Web
    • Client discovery
    • Strategy
    Visit site: GreenSpace Lawn Care

Open source

Everything on GitHub
  • NextGen_HydraPython

    End-to-end automation to acquire, verify, and tidy historical NOAA NextGen streamflow data.

  • hydra-nwm-streamflow-correctionPython

    GRU, Transformer, and conditioning-head models for National Water Model streamflow time-series regression.

  • Runoff_ForcastingPython

    Thesis-era pipeline: preprocess data, train a deep-learning model, evaluate the corrected forecasts.

  • PortfolioTypeScript

    This website: Next.js 16, React 19, Tailwind CSS 4, a Resend contact API, and a Calendly embed.

04 · Research & Talks

Research and speaking.

Where the work is being presented and published. Each item carries its current status and is updated as milestones land.

Full research page

  • December 2026

    Workshop

    Best Practices for AI and Agentic Workflows in Earth Science Research

    AGU26 Annual Meeting · San Francisco · December 7–11, 2026 · Scientific workshop facilitator

    Accepted

  • Decision pending

    Abstract

    HYDRA streamflow-forecasting abstract

    AGU26 · Hydrology session H100 (machine learning in hydrology)

    Under review

  • In preparation

    Manuscript

    HYDRA results manuscript

    Water Resources Research (target journal) · Author

    In progress

  • Planned

    Manuscript

    NextGen reforecast generation software

    Environmental Modelling & Software (target journal) · Author

    Planned

  • December 2025

    Thesis

    Senior Honors Thesis on runoff forecasting with deep learning

    Appalachian State University · Author

    Completed

05 · Skills

What I bring to a team.

Technical depth from research and industry, and operational habits from the Air Force.

  • Machine learning and data

    • PyTorch · MLX · scikit-learn
    • LSTM, Transformer, and Mamba-style models
    • Leakage-aware temporal evaluation
    • RMSE · NSE · KGE
    • pandas · NumPy · xarray · PyArrow
    • PostgreSQL · SQLite

    Used inHYDRA, Harmony, UT Austin coursework

  • Backend and web

    • Java · Spring Boot
    • GraphQL
    • FastAPI
    • React · Next.js
    • Node.js
    • HTML/CSS · Tailwind CSS

    Used inUSAA, Harmony, client work, this site

  • Cloud and tools

    • Google Cloud: BigQuery, Cloud Storage
    • AWS research compute (CPU/GPU)
    • Docker
    • Git · pytest · Jira

    Used inHYDRA, Harmony, USAA

  • Languages

    • Java
    • Python
    • JavaScript / TypeScript
    • SQL
    • C · C++ · Assembly
    • Swift · R

  • Operations and leadership

    • Safety-critical operations
    • In-flight service for distinguished visitors
    • Mission logistics across 30+ countries
    • Visas, crew hotels, and subsistence billing
    • Coordination with White House staff and security teams
    • Technical communication
    • Active TS/SCI clearance

    Used inU.S. Air Force (89th Airlift Wing, Air Force Two), USAA, AGU26 workshop

06 · Contact

Let’s talk.

I’m open to software engineering and applied AI roles, including scientific computing and mission-driven teams where an active clearance matters. I’m also glad to talk forecasting, evaluation, or the move from military service into AI.

mitchel.carson@gmail.com
Schedule a conversationRésumé (PDF)
LinkedIn
linkedin.com/in/mitchelcarson
GitHub
github.com/Mitchel34
Based in
Austin, Texas · Central Time
Clearance
Active TS/SCI

Prefer a form or an embedded calendar? Use the contact page.