// hello, world
A showcase of personal projects, professional work, and real-world problem solving.
View My WorkFull-featured personal AI system built on Claude Code. 7-phase execution engine, 49 skills, lifecycle hooks, persistent memory, and ElevenLabs voice interface — running on Bazzite Linux.
Fully automated deployment pipeline. Python file watcher + GitHub Actions + SSH deploy to Namecheap. Changes go live within seconds of a local edit.
Two Python CLI tools for converting PDFs and Word documents to LLM-ready Markdown. pdf2md handles books and batch jobs. paper2md handles multi-column research papers.
Built a catalog item that queries a banking API in real time — auto-populating customer name and account type as the user types, with automatic Help Desk routing on submission.
Designed and deployed a Claude-powered self-service chat bot inside ServiceNow — giving employees instant answers about their tickets, requests, and catalog items without opening a single form.
Took the AI chat widget from a working prototype to an enterprise-ready app — adding four role-based tiers, L1 troubleshooting flows, write actions, audit logging, rate limiting, and CIS-compliant credential storage.
The next leap in AI-augmented work isn't a better model. It's a better system around the model. Here's the architecture we built — and why Claude Haiku inside it outperforms raw Claude Opus on the same tasks.
Built and tuned a self-hosted LLM inference server on a single consumer GPU — nearly tripling its throughput with smarter model placement — then made it privately shareable with a team over an encrypted mesh network with key-based auth.
Problems I encountered in the real world — and how I solved them.
Problem: MCP connections to ServiceNow failed silently due to URL configuration, certificate trust, and SSO authentication — then once connected, needed to validate real write operations.
Fix: Resolved all three failure layers, established a live connection, and used it to create and configure actual ServiceNow records directly from the AI agent.
Problem: ServiceNow workspace displayed blank for regular users but worked for admins, with no error messages or clues.
Fix: Resolved variant audience matching and sort order precedence—admin testing masked the issue.
Problem: Catalog items appeared to submit successfully but never created RITMs—silent failure with no error messages.
Fix: Found onSubmit client script returning false, blocking submissions. Fixed return value semantics.
Problem: Users had to manually look up account details in external banking system when entering account numbers in ServiceNow.
Fix: Built auto-population using REST Messages, GlideAjax, and multi-type fallback to query banking API directly from catalog items.
Problem: Catalog items needed to auto-populate from an external API without a page reload, using OAuth2 and vendor-specific headers.
Fix: Four-object ServiceNow stack — REST Message, two Script Includes, GlideAjax client script — with multi-type account fallback and full debugging checklist.
sys_ui_script with global=true sounds like it should work everywhere. It doesn't. Here's the exact diagnostic, the correct injection pattern for Service Portal, and three other bugs we hit building a Claude-powered chat widget.
A security scan flagged missing HSTS, CSP, X-Frame-Options, and three other headers on a static site hosted on Namecheap Apache. Here's the exact .htaccess fix and why each header matters.
Infrastructure engineer and systems builder with 6+ years of experience working across banking, healthcare, and tech environments. Over time, my focus shifted from traditional infrastructure and automation into AI systems and practical agent workflows.
I'm especially interested in how AI can improve real operational work — not just demos or hype, but tools that actually save time, reduce friction, and help people solve problems faster. Most of the projects in this portfolio come from experimenting, building, and figuring things out hands-on.
Right now I'm focused on agentic AI, infrastructure for AI tools, automation pipelines, and integrating AI into existing enterprise systems. I enjoy bridging the gap between technical systems, APIs, automation, and user workflows to create things that are genuinely useful.
This site is a running collection of projects, experiments, and lessons learned as I continue exploring where AI fits into real-world infrastructure and operations.
Want to get in touch? Find me here.