// summary
AI engineering leader specializing in the design and security of production agentic systems. I architect agent runtimes, tool execution and sandboxing, memory and retrieval, orchestration, and evaluation, then build the guardrails and red-teaming that keep them reliable in adversarial, high-consequence environments. At HiddenLayer I secure agentic systems for enterprise customers and build product for agentic red teaming and adversarial prompt engineering. I founded cAIge, the vendor-agnostic Certified AI Guardrail Engineer program, defining an industry taxonomy for prompt-injection defense, data-exfiltration prevention, and guardrail validation. Deep working knowledge of foundation-model internals and the math beneath them, backed by a decade of architecture, engineering, and technical leadership across enterprise and education at scale.
// experience
AI Architect
HiddenLayer
Oct 2025 – Present
- ▸Architect and secure production agentic systems for enterprise customers: agent loops, tool execution and sandboxing, memory and retrieval, and security boundaries, threat-modeling prompt injection, tool-output poisoning, and sandbox escape
- ▸Build product for agentic red teaming and adversarial prompt engineering, generating multi-turn attack flows that surface misuse and failure modes in customer AI systems
- ▸Develop core platform capabilities and customer integrations, turning research prototypes into reliable, shipped product
- ▸Advise customers on evaluation and guardrail strategy grounded in foundation-model internals, balancing operator trust, reliability, and performance
Chief Research Scientist
Average Joes Lab
Jul 2025 – Present
- ▸Founded an open, community-driven research lab and lead its applied AI work: agent runtimes, episodic and semantic memory harnesses, sandboxed tool execution, and multi-provider inference gateways
- ▸Drove autonomous-research programs to systematically discover better context-management, memory, tool-use, planning, and orchestration strategies for agents
- ▸Built adversarial-AI research, including a generalized red-team agent and a zero-training, embedding-space prompt-injection detector
- ▸Authored a framework-free, code-first agentic-engineering curriculum and a research-engineering learning path, mentoring newcomers into independent research
Chief Enterprise Architect
Odovey Consulting
Jun 2025 – Present
- ▸Lead AI and cloud strategy for enterprise clients: AI agent development, RAG implementation, supervised fine-tuning, and AI model infrastructure
- ▸Authored a vendor-agnostic GenAI adoption framework for planning, governing, and scaling generative AI across the enterprise
- ▸Advise executives on AI architecture and adoption, aligning AI investments with security and measurable business outcomes
Senior Content Developer
Microsoft
Aug 2024 – Oct 2025
- ▸Spearheaded technical enablement for Azure Container Registry (ACR) on Microsoft Docs, integrating AI-driven workflows with cloud architecture expertise
- ▸Designed and published scenario-driven content, reaching over 500K developers monthly
- ▸Developed an AI-powered authoring pipeline, reducing time-to-first-draft from 3 days to under 5 minutes
Chief Executive Officer
Chasing Cloud Careers
Aug 2023 – Present
- ▸Co-founded Chasing Cloud Careers, focusing on accessible engineering education
- ▸Curated free learning resources into structured paths for various engineering roles
- ▸Mentored professionals in career transitions, fostering a supportive learning community
- ▸Developed self-directed learning approaches that empower individuals to explore technology careers
Principal Cloud Architect
CBTS
Sep 2021 – Jul 2024
- ▸Designed and implemented secure, scalable cloud solutions on Azure and AWS for migrations, modernization, and AI-enabled architecture
- ▸Delivered 13 enterprise projects, averaging $94K per engagement, focusing on AI/LLM-driven architectures and enterprise AI adoption
- ▸Designed cloud-native PoCs using Hugging Face NLP models and advised executives on scaling AI workloads across multi-cloud platforms
- ▸Engineered reusable IaC patterns, reducing deployment errors by 40% and provisioning effort significantly
Training Architect - Azure
Linux Academy | A Cloud Guru | Pluralsight
Jun 2019 – Aug 2024
- ▸Authored 9 Azure certification courses, achieving over 20.8M minutes viewed and a 4.1/5 learner rating
- ▸Developed 50+ auto-provisioned labs, enabling 175K+ completions while reducing setup time by 90%
- ▸Audited over 150 hours of content, resolving ~1,300 tickets and improving course ratings from 4.3 to 4.6
- ▸Authored and refreshed 400+ exam assets, boosting learner pass rates by 10% for key certifications
Jr. Systems Administrator
Endurance International Group
Sep 2018 – Jun 2019
- ▸Supported shared, VPS, and dedicated hosting environments for over 600 SMB customers in a 24/7 operations team
- ▸Resolved ~105 support tickets monthly with a 96% customer satisfaction rate
- ▸Automated repetitive tasks using Bash, reducing average ticket handle time by 18%
- ▸Executed zero-downtime migrations for 80+ WordPress sites, ensuring SLA compliance
// selected_projects
cAIge: Certified AI Guardrail Engineer
Founded the vendor-agnostic AI guardrail engineering certification: a five-domain taxonomy and exam covering prompt-injection defense, data-exfiltration prevention, adversarial attack mitigation, and guardrail validation.
caige.orgAgent runtimes and orchestration
Standalone agent execution engines, the loop, tool dispatch, and event streaming, plus orchestration patterns (routing, orchestrator-workers, evaluator-optimizer) and a code-first agentic-engineering curriculum.
Sandboxed tool execution
Isolated Docker execution for agent tools with seccomp, dropped capabilities, resource limits, egress control, and lifecycle management to contain prompt-injection blast radius.
Agent memory harnesses
Episodic and semantic memory for agents: write-time fact extraction, hybrid BM25-plus-vector retrieval, and token-accurate working-memory management.
Adversarial AI and red teaming
A generalized red-team agent and a zero-training, embedding-space prompt-injection detector for evaluating and hardening agentic systems.
Inference gateways and engines
Multi-provider LLM gateways with routing, fallbacks, PII guardrails, quotas, and observability, plus an LLM inference engine and runtime built from scratch in Rust, Go, and PyTorch.
Technical writing: cdovey.dev
A deep technical series on LLM internals, agent architecture, and AI security: agent loops, memory, sandboxing, prompt injection, red teaming, and evaluation.
cdovey.dev/blog// certifications
Microsoft Azure
- ▸Azure Solutions Architect Expert
- ▸Azure Administrator Associate
- ▸Azure Security Engineer Associate
- ▸Azure Network Engineer Associate
- ▸Azure AI Fundamentals
- ▸Azure Fundamentals
- ▸Microsoft Certified Trainer
AWS
- ▸AWS Solutions Architect Associate
- ▸AWS Cloud Practitioner
Security
- ▸CompTIA Security+
- ▸CompTIA CySA+
- ▸CompTIA Security Analytics Professional (CSAP)
Other
- ▸PCEP – Certified Entry-Level Python Programmer
- ▸Linux Essentials (LPI)
- ▸Intro to Gen AI Mini Course — AI by Hand (1st Cohort)
// volunteer
Robotics Mentor
The Awty International School
Aug 2025 – Present
Mentoring students on programming and electrical engineering for FIRST Robotics Competition for FIRST Championship.