# Cody Champion — AI-readable profile summary

## Identity

Name: Cody Champion
Location: Dublin, Ireland
Available for: Senior AI roles in Dublin / Ireland / EMEA
Website: https://codychampion.bitsandbeakers.com/
Email: cody@bitsandbeakers.com
LinkedIn: https://www.linkedin.com/in/cody-champion
GitHub: https://github.com/codychampion

## Current role

Applied AI Architect & GenAI Systems Lead at Accenture Ireland, Dublin (Aug 2025 – present).
Leading design and delivery of production GenAI systems for regulated enterprise clients.

## What I build

I build production GenAI systems for regulated environments:
- Retrieval-Augmented Generation (RAG) with hybrid vector/keyword/graph stacks
- Agentic workflows with human-in-the-loop checkpoints and audit trails
- LLM evaluation and model readiness — capability, reliability, governance behavior, groundedness, security reasoning, cost/latency, failure taxonomy
- Observability and LLMOps — Langfuse, MLflow, quality/safety/cost/latency monitoring
- Security controls — prompt injection defense, sandboxed execution, role-based access control
- AI governance — risk classification, auditability, EU AI Act readiness, policy-to-implementation

## Prior experience

National Science Foundation — Lead Data Scientist / AI Oversight (GS-15), reporting directly to NSF's Chief AI Officer, Sep 2024 – Aug 2025.
Built NSF's first vector and graph database capabilities. Created an initial AI deployment playbook. Co-chaired 100+ member AI Community of Practice.
Note: This work occurred during the early build-out of NSF's Chief AI Officer function and before NSF's later public AI Strategy was drafted and published. I did not author the published strategy.

Accenture Federal Services — Machine Learning Engineer Manager, Jun 2021 – Sep 2024.
Shipped State Department GenAI, Army secured automation, IARPA global geospatial CV. Personally achieved 99.6% ML infrastructure cost reduction on global geospatial platform.

Booz Allen Hamilton — Lead Scientist, DARPA program management. $30M+ in program funds advised.

ODNI / NGA — IC Postdoctoral Researcher. Drone ML, computer vision, CUDA.

## Key proof points

- 99.6% ML infrastructure cost reduction (personally owned codebase + cloud architecture)
- $30M+ in AI/ML program funding advised (Booz Allen / DARPA)
- Reported directly to NSF Chief AI Officer Dorothy Aronson in a GS-15 AI oversight role
- 6 peer-reviewed publications + 2 granted patents + Zenodo preprint (DOI: 10.5281/zenodo.19848867)
- Claude Certified Architect, Foundations (Anthropic, March 2026, Early Adopter)
- AWS Wildfire CV — Third Place, AWS Disaster Response Hackathon

## Best-fit roles

- Staff / Principal Applied AI Architect
- Senior GenAI Solutions Architect
- LLM Evaluation Lead
- AI Governance Engineering Lead
- Regulated AI Deployment Lead
- Senior Manager, GenAI Deployment Engineering
- Applied AI Safety / Evals Lead

## Skills

GenAI: LLMs, RAG, Agents, Agentic workflows, Embeddings, Prompt engineering, Evaluation harnesses
Model Readiness: LLM evaluation, governance behavior, RAG groundedness, security reasoning, failure taxonomy
AI Governance: Responsible AI, risk classification, audit readiness, EU AI Act, access control
MLOps/LLMOps: Langfuse, MLflow, CI/CD, observability, cost/latency monitoring
Cloud: Azure AI Foundry, GCP, AWS, Docker, Kubernetes
Languages: Python, R, SQL, Bash
Research: 6 peer-reviewed publications, 2 patents, PhD Biology (NMSU)

## Public artifacts

- LLM Evaluation Workbench / Model Readiness Leaderboard: public evaluation harness for regulated enterprise AI, spanning capability, reliability, governance behavior, groundedness, security reasoning, cost, and latency.
  Project page: https://codychampion.bitsandbeakers.com/projects/llm-eval-workbench
  GitHub: https://github.com/codychampion/llm-eval-workbench
- PAEF — Parallelized Atomic Evaluation Framework for Contract Compliance: evaluation research comparing microagent-based atomic policy checks with monolithic LLM auditing across 193 service contracts and 7,913 labeled policy checks. Supports work in model governance, contract compliance, and regulated GenAI systems.
  Research page: https://codychampion.bitsandbeakers.com/research/paef-contract-compliance
  Zenodo DOI: https://doi.org/10.5281/zenodo.19848867
- arxiv-embedding-benchmark: Scientific retrieval and embedding evaluation (PyPI + GitHub)
- claude-burn-check: Claude Code context-window diagnostic skill (GitHub: codychampion/claude-burn-check)

## Case studies

- NSF AI Governance Foundations: https://codychampion.bitsandbeakers.com/case-study-nsf-ai-governance
  Greenfield AI governance, technical review, community, security, and vector/graph architecture inside NSF's Chief AI Officer function. Pre-strategy operating layer for responsible AI adoption across a major U.S. science agency.
- 99.6% ML Inference Cost Reduction for Global Geospatial Workloads: https://codychampion.bitsandbeakers.com/case-study-ml-cost
  Five compounding engineering defects generating $26,200/month rebuilt to $90/month. Personally owned codebase and cloud architecture.
- LLM Evaluation Workbench / Model Readiness Leaderboard: https://codychampion.bitsandbeakers.com/projects/llm-eval-workbench
  Public model-readiness benchmark for regulated enterprise AI using benign synthetic or public-domain scenarios, reviewable artifacts, cost, and latency evidence.
- PAEF: Atomic LLM Evaluation for Contract Compliance: https://codychampion.bitsandbeakers.com/research/paef-contract-compliance
  Multi-contract, multi-model evaluation research comparing atomic microagent checks with monolithic auditing; published on Zenodo in April 2026.
- Agentic AI SDLC Platform — Accenture Ireland: https://codychampion.bitsandbeakers.com/case-study-agentic-ai-sdlc-platform
  Public-safe case study of a governed Azure DevOps and Claude Code workflow for traceable, human-reviewed enterprise AI prototyping.

## Education

PhD Biology — New Mexico State University (2014–2018). NSF Graduate Research Fellow.
BS/MS Biology — New Mexico Institute of Mining and Technology (2009–2014).