Codecademy AI Maker & Agentic AI graduate
Sr Network Engineer & AI Builder
I've spent years automating contact-center systems—cutting processes that took hours down to seconds. After two Codecademy programs in agentic AI, I'm applying that same instinct to multi-agent systems, RAG pipelines, and tools built with CrewAI, LangGraph, and Claude.
Selected work
From retrieval systems to agent orchestration—each build starts with a real workflow pain point and ends with something people actually want to open every morning. Hover over a project's tile to see more screenshots.
Problem → Analysts had to stitch together market data, filings, and news by hand before they could form a thesis. Solution → Built a multi-agent CrewAI system with hybrid RAG that researches, cross-checks, and drafts financial analysis through a FastAPI backend and Streamlit UI.
View on GitHub
Problem → Off-the-shelf voice assistants don't let you choose your own LLM or add custom skills. Solution → Built a voice/text assistant powered by Grok (xAI), using LangGraph for orchestration and Whisper for speech-to-text, wrapped in a Streamlit interface.
View on GitHub
Problem → Every AI agent that needs to do research ends up rebuilding its own fetch-scope-synthesize pipeline from scratch. Solution → Built a stateful MCP server with five coordinating subagents (Scope, Source, Conflict, Gap, Synthesis) that any agent can call as a shared research brain, with built-in conflict detection and context trimming.
View on GitHub
Problem → Market signals for equities, commodities, real estate, and crypto live in scattered dashboards with no narrative tying them together. Solution → Paired a Claude-powered reporting agent with Streamlit dashboards that compute trend, momentum, and valuation signals, then export the results as PDFs or Slack-ready reports.
View on GitHub
Problem → Getting quick, cited answers out of a public webpage usually means reading the whole thing yourself. Solution → Built a Streamlit tool powered by Claude that holds a multi-turn conversation grounded in a given webpage's content.
View on GitHub
Problem → Turning a meeting recording or shared audio link into usable notes means manual transcription. Solution → Built a tool that transcribes audio directly from a URL, cutting out the download-then-transcribe busywork.
View on GitHubExperience
Day-to-day, I design and automate the contact-center systems that keep customer-facing teams running—the same eye for turning slow, manual processes into something fast and reliable that I now bring to AI projects.
Introduced to the platform and, within about a week, designed workflows replicating Avaya functionality—call queues, role-based profiles, reporting, and call recording. Became the team's go-to resource for the platform.
Led the migration to realize annual cost savings and strengthen ties with Triton WFM.
Wrote a script to retire 300+ DRC/DRTV toll-free numbers, cutting a batch process from hours to seconds.
Implemented a single global Flash Announcement variable across six workflows for consistent configuration and lower maintenance.
Designed two Avaya reports for call-center visibility, and built Python automation to add international offices and streamline ServiceNow ticket reassignment.