DTRA · PROBANKER

The Problem

DTRA needed a training environment where analysts and operators could practice recognizing and countering adversary information operations against evolving, realistic scenarios. Static doctrine couldn't adapt to trainee decisions or produce the nuanced critique required to build real-world judgment under information-warfare conditions.

What I Built

• Integrated a Llama-3 8B critique engine that evaluates trainee responses in real time across adversary intent, propaganda technique, psychological levers, and countermeasure selection

• Built the trainee-facing Angular 20 frontend — scenario presentation, response capture, and AI-critique rendering — wired to the Django API

• Built the DDAY simulator as a plug-in module inside ProBanker's multi-database Django architecture — scenario state, adversary logic, and LLM-scored feedback routed per course

• Stood up a dedicated DTRA deployment environment on AWS, segregated from the commercial ProBanker footprint, with its own settings, CI/CD, and secrets

• Translated DTRA scenario doctrine into structured game data with subject-matter experts — scenarios, question banks, and scoring rubrics the LLM grounds its critique against

Outcomes