Available for senior AI & platform engineering leadership roles
Sr. Director of AI Engineering · DevSecOps · Agentic AI · Dallas, TX
15+ years leading DevOps platforms at scale, and most recently shipping production Agentic AI — a multi-skill agent for Terraform automation that runs 3–5× faster than ChatGPT and Claude on identical tasks.
Two production-grade multi-agent systems, built in the open, MIT licensed. One responds to pipeline failures in real time. The other learns from them afterwards. Click either demo — no signup, no API key.
Autonomous CI/CD incident response — AI agents that triage failures and open fix PRs while you sleep.
Four coordinated agents (Watcher → Diagnoser → Fixer → Notifier) compress the typical 30–90 minute 2 AM triage cycle into minutes. Detects the failure, pulls logs, diagnoses root cause with Claude, commits a fix to a branch, opens a draft PR, and notifies the team on Slack. Docker Compose ready; dry-run mode for safe onboarding.
Autonomous incident post-mortem engine — reconstructs timelines, runs RCA, builds a learning knowledge base.
Hierarchical multi-agent system: an Orchestrator coordinates six specialists (evidence, timeline, RCA, action items, writer, evaluator). Every draft is scored by an LLM-as-judge against a 5-dimension rubric — low scores trigger a revision loop, bounded to 3 cycles to prevent runaway. Approved post-mortems are embedded and stored in ChromaDB so future incidents learn from past ones. Typed Pydantic contracts between every agent boundary.
"We're excited to welcome Adnan to GDT's team of talented, forward-thinking IT engineers and professionals." — Brad Davenport, VP of Solutions Engineering, GDTRead the full announcement ↗
Deep-dives on the AI systems I've built — architecture, tradeoffs, and what the future of DevOps looks like when agents do the mechanical work.
Why tribal knowledge disappears when engineers leave — and how retro-pilot turns every incident into a permanent, searchable lesson for the next team that hits the same problem.
Read on LinkedIn ↗The architecture behind ops-pilot: how four coordinated agents detect failures, diagnose root cause, write the fix, and open a draft PR — without removing humans from the review loop. A case for agents as force multipliers, not replacements.
Read on LinkedIn ↗The DuploCloud system in detail: 11 specialized skills, async execution on Kubernetes, NATS JetStream, AWS Bedrock. Built to remove humans from the execution loop — while keeping the review gate where it matters.
Read on LinkedIn ↗Agentic AI · Multi-Agent Orchestration · LLM-as-Judge · Prompt Engineering · Prompt Caching · Claude (Bedrock) · ChromaDB / Vector Stores · Pydantic · Python
Terraform · Terragrunt · Kubernetes · Helm · CI/CD · Jenkins · GitHub Actions · Docker
AWS · Hybrid Cloud · EKS · EFS · NATS JetStream · SSE · DevSecOps · OWASP · SRE
Team building (2 → 15+) · 0 → 1 platform builds · Cross-functional leadership · Hiring & mentoring
I'm open to senior AI engineering, platform engineering, and DevOps leadership roles — remote, or hybrid in the Dallas, TX area.