A working portfolio of systems, prototypes, and modernization concepts spanning AI, cybersecurity, autonomous tooling, and software delivery.

AI Selection Board

A structured evaluation system for comparing AI models, vendors, or implementation options against mission, security, and delivery constraints.

  • Problem: AI selection decisions often become ad hoc, difficult to reproduce, and hard to explain to technical, acquisition, and leadership audiences.
  • Approach: Build a repeatable board process with scored criteria, narrative justification, evaluation artifacts, and decision records that can be revisited as the market changes.
  • Current status: Active concept

AI-Assisted Software Development Tools

A set of patterns, utilities, and process ideas for applying AI responsibly across specification, coding, review, and delivery work.

  • Problem: AI development workflows can speed up output while also creating inconsistency, shallow review, and governance concerns if they are introduced without structure.
  • Approach: Focus on practical use cases such as code generation with review gates, repo-aware planning, evaluation harnesses, and documentation support that fit real engineering workflows.
  • Current status: Active exploration

Agentic News and Research System

A research intake and synthesis workflow for monitoring technical developments, grouping related signals, and producing concise briefs.

  • Problem: Tracking fast-moving AI, security, and tooling developments by hand is time-consuming and often produces fragmented understanding instead of durable research insight.
  • Approach: Combine curated sources, lightweight automation, extraction pipelines, and synthesis prompts to move from raw inputs to categorized briefs and reusable research notes.
  • Current status: Ongoing

Autonomous Penetration Testing Platform

A concept and prototype track for guided offensive assessment workflows that combine reasoning, tooling, operator checkpoints, and reporting.

  • Problem: Traditional penetration testing automation is good at running tools, but weak at chaining context, adapting to findings, and producing usable outputs for operators and stakeholders.
  • Approach: Explore an agentic workflow that can plan scoped assessments, select tools, interpret intermediate results, request approval at key decision points, and convert findings into structured evidence and remediation notes.
  • Current status: Concept development

CyberTabletop / Incident Response Tabletop Agent

A tabletop facilitation concept for guiding response scenarios, prompting decision points, and preserving exercise outputs for after-action review.

  • Problem: Tabletop exercises often rely on manual facilitation, inconsistent prompts, and notes that are difficult to convert into repeatable learning or operational improvement.
  • Approach: Use an AI-assisted facilitation workflow to present scenario branches, inject prompts, adapt follow-up questions to participant choices, and produce structured after-action materials.
  • Current status: Prototype direction

Federal ATO Reform / Agentic ATO Concepts

A modernization concept focused on reducing friction in federal authorization workflows through automation, evidence reuse, and agent-assisted coordination.

  • Problem: Authorization processes frequently involve manual evidence collection, fragmented ownership, and duplicated effort that slows secure delivery.
  • Approach: Explore how agentic support, evidence mapping, and workflow automation could streamline collaboration between delivery teams, security stakeholders, and authorizing roles.
  • Current status: Research