About Me

Posted on Aug 24, 2026

Software engineer at IBM, based in Raleigh, NC. I work on agentic AI for IBM Z and z/OS.

About

A lot of my job is working out where AI genuinely helps. These tools make it easy to build software quickly and just as easy to stop thinking about it — that second part is where things quietly go wrong. I’d rather use them in a way that leaves engineers sharper.

Two things I keep coming back to. First, fundamentals: AI helps most once you can tell whether its output is any good, and that judgment comes from having done the work yourself. Second, reuse over rebuild — now that anyone can generate software, the same thing tends to get built five times instead of once.

Right now I’m on the z Platform Evaluation Team (zPET). I install z/OS AI products, run them hard, find where customers will struggle, and add AI and automation so our experts can spend their time on the harder problems. I also help steer AI adoption across IBM Infrastructure test teams.

For fun, I lift weights, train Brazilian Jiu-Jitsu, read books, and play games.

Journey

I joined IBM in 2020 as an intern while finishing my Computer Science degree at Siena College, worked on the z/TPF team through graduation, and came on full time in 2021.

Since then I’ve moved through software engineering, ML, and architecture roles, all pointed at z/OS and bringing AI to IBM Z. I still write plenty of code — RAG pipelines and ingestion tooling for generative AI — and spend a fair amount of time translating between the engineers and the people who sign off on the work.

Experience

IBM — Software Engineer, z Platform Evaluation Team (zPET) (Oct 2024–Present)

Work with first- and second-line managers across IBM Infrastructure test teams (Storage, Power, Z, Linux) on AI-transformation strategy and KPIs: as AI coding agents make duplicated effort cheap, I set reuse-and-contribute standards and track cross-team tool adoption so we build on shared code instead of reinventing it.
Install, administer, and stress-test agentic AI for z/OS (Machine Learning for z/OS, AI System Services, watsonx Assistant for Z) on OpenShift and Cloud Pak for Data to surface customer-specific problems; filed dozens of defects and coached developers on running containerized agentic services, including Spyre and the Polaris z/OS modernization effort.
Lead AI and automation work that frees subject-matter experts from repetitive tasks; led a cross-functional team upskilling z/OS engineers to build agentic solutions for Dump Management.
Built multi-modal RAG pipelines ingesting video, PDFs, runlogs, and GitHub wikis to power internal automation; cut manual regression-testing costs ~88% via Galasa, Ansible, and Jenkins.

IBM — Technical Architect (Dec 2023–Oct 2024)

Built AskiCDNet, a generative-AI RAG assistant for IBM’s Client Delivery Network (LangChain, Milvus, watsonx.ai / watsonx.data), and led its migration from a Java monolith to Python microservices; mentored the team on architecture patterns.

IBM — ML Engineer (Dec 2022–Dec 2023)

Built CI/CD pipelines and a Java Spring Boot API integrating watsonx Assistant, ServiceNow, and Palo Alto Prisma; cut request latency 40% through caching and query optimization.

IBM — Software Engineer (Jul 2021–Dec 2022)

Shipped Node.js REST APIs for z/TPF systems and built a VS Code extension enabling remote z/TPF C/C++ development.

Home Lab

I run a self-hosted, two-node AI inference cluster at home, all managed with Ansible. One box does inference — quantized GGUF models up to 35B across an AMD GPU and NPU with llama.cpp — and the other runs orchestration (a litellm proxy), Grafana/Prometheus, a Tailscale TLS mesh, and a kernel-isolated agent sandbox. It’s where I learn inference internals for real: quantization, KV cache, GPU vs NPU, and where things break under agent traffic. I write up what I learn on my blog.

Speaking & Writing

I speak regularly — small workshops up to 500+ people — on spec-driven development, agentic workflows, responsible AI, and inner-/open-source (including Hacktoberfest). After those sessions my team moved from ad-hoc “vibe coding” to spec-first development, which is how Neo3270 got built. I write on my personal blog and the zPET community blog about agentic engineering, token-efficient LLM context, IBM Z modernization, and self-hosting quantized models.

Skills

Agentic AI: Building agents, skills, and MCP servers/tools; multi-agent systems; RAG; LLM orchestration; spec-driven development Languages: Python, Go, Java, SQL Platforms: IBM Z / z/OS, OpenShift, Cloud Pak for Data, IBM Cloud, Kubernetes, Docker

Certifications

IBM Developer Profession, Level 2 (Experienced), 2026
IBM Consulting Generative AI Foundations
IBM Agile Explorer

Innovation

I’m a named inventor on multiple IBM patents, with further invention disclosures filed. I also help other engineers work through the disclosure process — turning good ideas into something protected and real.