I identify broken or manual systems, figure out what's actually wrong, and build the tools that give decision makers the information they need to act.
The ASOS program — critical federal weather observation infrastructure supporting NOAA, FAA, and DOD nationwide — needed a more robust, data-driven approach to procurement planning. I built a comprehensive framework from scratch, designing both the data layer and the front-end tools that non-technical logisticians and leadership use to guide purchasing decisions. The result: backorders decreased, zero-balance items were reduced, and more targeted purchasing decisions reduced overall spend while ensuring critical items stayed in stock.
A legacy financial database was decommissioned. The only surviving copy of multi-million dollar budget and contract data — covering key federal projects and active contracts — was locked inside 1,800+ pages of static PDFs. With no established process for recovering the data, preserving the integrity of those financial records required building a solution from scratch.
I designed an algorithmic extraction engine using Power Query (Excel's automation language) to parse the PDFs and reconstruct the database programmatically, then built a verification layer to audit every record against the source.
The NWS Logistics Management Branch had an opportunity to modernize its reporting infrastructure — transitioning from a legacy PDF-based workflow to a more standardized, maintainable system. I led that modernization effort across six federal stakeholder groups — including ASOS, NEXRAD, and the National Reconditioning Center.
I transitioned all reports from PDF deliverables to structured Excel reports, then audited and updated the underlying SQL scripts for accuracy and maintainability. To accelerate that process, I built a custom AI tool using Google Gemini that ingests any SQL script and outputs an executive summary, inline comments explaining each section, and a technical risk assessment — turning hours of manual review into minutes.
To make the new system sustainable long-term, I built a shared cloud repository on Google Drive housing all updated SQL scripts, report templates, and SOPs I authored for report delivery. I also created and shared a Google Calendar with the data team mapping every report to its delivery deadline — centralizing the delivery process in a single shared source of truth.
LMB branch leadership wanted greater visibility into how budget was being allocated across programs — specifically the ability to track NEXRAD and ASOS spend separately rather than as a combined figure. I designed a custom SQL query to aggregate all monthly customer requisitions, factor in purchases already made by LMB staff, and feed a front-end dashboard that gives leadership a real-time view of available balance by appropriation code — segmented by program for the first time.
Potencia is a nonprofit serving adult immigrants to build economic mobility. Their team was spending significant time on repetitive marketing and fundraising tasks without a scalable knowledge system behind them.
I designed and deployed a custom GPT with the organization's knowledge, voice, and program details built in — delivered with a full prompt library, usage documentation, and staff training. The team now uses it actively for marketing content, donor communications, and grant writing support.
I wanted to understand what it actually takes to build a production AI product — not follow a tutorial, but own every decision. So I built one from scratch.
I built a production-grade AI search system handling everything from how documents get processed and indexed, to how queries get matched and ranked, to how accuracy gets measured — all running on cloud infrastructure. An ongoing personal project I continue to develop.
Every role I've had has had the same shape — walk into something broken or manual, figure out what's actually wrong, and build something that gives the right people the information they need to act. I've done that inside large, complex federal organizations where the stakes are real — which means I know how to move through ambiguity, work across teams, and ship things that actually get used.
My background is unusual for someone in this space: Psychology and Entrepreneurship at Tufts, then self-taught in data engineering, SQL, and AI. Most people come from pure engineering or pure strategy. I come from a different angle — I've always been more interested in how people actually think and make decisions than in the technology itself.
That background shows up in everything I build. When I designed the procurement lookup tool, I wasn't thinking about the data model first — I was thinking about a non-technical team member under pressure who needs one answer, fast. When I present to leadership, I'm not reciting metrics — I'm translating complexity into the specific framing that drives a decision. And when I'm working across engineering, logistics, and executive teams at the same time, I'm not just coordinating — I'm reading each group, adapting how I communicate, and pushing the work forward across people who often want very different things.
That combination — systems thinking, technical building, and human-centered design — is what I bring to every problem. It's why the tools I build actually get used.
I led a 20-person student marketing agency at Tufts. I TA'd an entrepreneurship and startup design course. I've advised a nonprofit on AI strategy. I've rebuilt federal data infrastructure from scratch with no team and no playbook. The common thread is ownership — I walk in, I see the whole system, and I make it better.