Hi, I'm David Plehn

~/now

What I'm doing now

1. Research Assistant

   ACTIVE

   Bayesian Cognitive Modeling Lab (Prof. Michael D. Lee), UC Irvine

   Irvine, CA · Oct 2026 – Present

   Building a Bayesian Rasch latent-truth model in JAGS that weights wisdom-of-the-crowd judgments by each person's inferred expertise, and extending it to per-topic ability and difficulty parameters.

2. Research Assistant

   ACTIVE

   Learning and Decision Neuroscience Lab (Prof. Mimi Liljeholm), UC Irvine

   Irvine, CA · Oct 2026 – Present

   Deep reinforcement learning project on social learning — observational Q-learning agents comparing imitation with emulation, and MiniGrid agents testing instrumental control as an intrinsic reward.

3. Systems & Software Engineer Intern

   ACTIVE

   Lockheed Martin

   Colorado Springs, CO · Jun 2026 – Present

   Designed and deployed an enterprise-wide RAG AI chatbot for the C2BMC program, integrated AWS Bedrock AI tooling into GitLab CI/CD pipelines, and automated Cameo MBSE projects, Ansible playbooks, and a ContainerLab virtual network environment for mission-critical systems.

4. Artificial Intelligence Trainer

   ACTIVE

   Handshake AI

   Irvine, CA · Nov 2025 – Present

   Training and evaluating multimodal AI systems across text, audio, and vision — domain-specific prompts, scoring rubrics, and AutoCAD training data that lifted cross-modal accuracy by roughly 15%.

~/projects --featured

SceneCheck

COMPLETE

Cross-platform event discovery app pairing location-based search with social features

• React Native
• Expo
• TypeScript
• Supabase
• +4

Full-Stack · 2026 read more

FinSight AI

WIP

Multi-agent platform analyzing global markets across technical, fundamental, sentiment, and macro signals

• Next.js
• FastAPI
• LangGraph
• Python
• +4

AI/ML · 2025 read more

GPT from Scratch

WIP

Transformer-based language model implemented from first principles

• PyTorch
• Python
• Transformers
• BPE Tokenization
• +3

AI/ML · 2026 read more

ML Wine Quality Predictor

COMPLETE

Multi-class wine quality prediction with logistic regression and feature analysis

• Python
• Scikit-learn
• Pandas
• NumPy
• +2

Data Science · 2024 read more

~/research

Research

1. Research Assistant — Bayesian Cognitive Modeling Lab

   ACTIVE

   Prof. Michael D. Lee · UC Irvine

   Irvine, CA · Oct 2026 – Present

   Bayesian cognitive modeling of expertise to improve wisdom-of-the-crowd aggregation.

   • Building a Bayesian Rasch (ability/difficulty) latent-truth model in JAGS that weights individual judgments by each person's inferred expertise, using the Bennett, Benjamin, Mistry & Steyvers (2018) dataset
   • Extending the model from a single holistic expertise parameter to per-topic ability and difficulty parameters so it can represent multidimensional expertise across knowledge domains
   • Surveying datasets in the lab's behavioral data repository to determine how existing cognitive models must be adapted to capture domain-specific expertise
2. Research Assistant — Learning and Decision Neuroscience Lab

   ACTIVE

   Prof. Mimi Liljeholm · UC Irvine

   Irvine, CA · Oct 2026 – Present

   Deep reinforcement learning models of observational and social learning, part of the lab's Agentic AI Experiment.

   • Combining the neural model of observational learning in Liljeholm, Molloy & O'Doherty (2012, J Neurosci) with the imitation-based social conformity model in Mauter & Liljeholm (2026, Brain and Behavior) into a single deep RL project
   • Implementing observational Q-learning agents to compare imitation (copying actions) with emulation (inferring others' goals) as mechanisms of social learning
   • Training RL agents in MiniGrid environments to test instrumental control (outcome divergence) as an intrinsic reward surrogate
3. AI Evaluation Research

   ACTIVE

   Handshake AI

   Irvine, CA · Nov 2025 – Present

   Training and evaluating multimodal AI systems across text, audio, and visual inputs.

   • Trained and evaluated multimodal AI systems across text, audio, and visual inputs, increasing response quality and cross-modal accuracy by ~15%
   • Developed and tested domain-specific prompts to benchmark large language model performance in scientific and technical subjects
   • Built detailed scoring rubrics to evaluate AI model outputs, increasing precision and consistency of evaluations across multiple AI labs
   • Designed AutoCAD Mechanical training data through multiple weeklong evaluation projects to improve model accuracy on professional CAD workflows
4. Research Assistant — Spintronics Lab

   PUBLISHED

   University of Denver

   Denver, CO · Oct 2024 – Jul 2025

   Experimental condensed-matter physics research on spin-orbit torque measurement, culminating in a co-authored peer-reviewed paper.

   • Developed LabVIEW programs for precise control of sensitive measurement equipment in an active physics research environment
   • Collaborated on AutoCAD designs to build custom tools enhancing lab measurement capabilities and reducing manual oversight by ~90%
   • Fabricated and validated ~10nm thin-film devices using vacuum sputtering, lithography, and related fabrication techniques for state-of-the-art MRAM research

   Co-authored “Field-like torque measured via a Wheatstone bridge” in the Journal of Applied Physics (2025)

~/publications

Publications

~/honors

Awards & honors

• [*]

  1st Place, MIC2 Intern Project Competition

  Lockheed Martin

• [*]

  2nd Place, CodeQuest Competition

  Lockheed Martin

• [*]

  Eagle Scout

  Boy Scouts of America

• [*]

  1st Degree Black Belt

  Karate