Case Studies, Real Systems, Real Numbers

Full-stack platform

Quote Manager

Custom freight quoting platform, built in a weekend

3.3×

win-rate lift

Challenge: Legacy quoting system on a Shanghai VM ($5,400/yr). Slow, disconnected from the CRM, invisible to leadership.
Solution: Built a full replacement over a weekend, Django + Next.js, deployed to the edge. Analyzed 33,992 quotes and found that answering within 16 hours wins about 3.3x as often.
Result: Generates more annual profit value than my salary. Eliminated the Shanghai VM, and leadership has self-service visibility for the first time.

A closer look

The finding that changed how the company quotes: a quote answered in under 16 hours wins about 3.3 times as often as one that takes longer. I pulled 33,992 real quotes to prove it, then built the platform to make fast the default.

It replaced a slow legacy tool running on a $5,400-a-year overseas VM. Rebuilt from scratch over a weekend with Django and Next.js, deployed to the edge, finally connected to the CRM, and visible to leadership for the first time.

Django REST APINext.jsPostgreSQLCargoWiseHubSpot

Data platform

Atlas

Turning a closed freight ERP into an open, queryable platform

Built solo

end to end

Challenge: The core operational data lived inside a closed freight ERP with no clean way to get it out. Reporting meant exporting to spreadsheets by hand. The CRM and the ERP told different stories. Every integration was bespoke and brittle.
Solution: Atlas is an API-first data platform that sits on top of the closed ERP and makes it behave like an open one. It ingests the data in real time and rebuilds history from a read-only replica, lands it in a clean canonical model, and exposes it through a secured REST API that feeds reporting, CRM sync, commission math, and an AI assistant, all on one foundation.
Result: The integration foundation the rest of the operation runs on. Phase one is live and on track to absorb a $40K/yr third-party visibility vendor as later modules ship. The part I'm proud of isn't any single piece, it's that the pieces fit together and stay correct, which is the hard part of a platform one person can maintain.

What's inside

Ingestion: a real-time event feed plus historical reconstruction from a read-only replica. Field mappings live in one declarative place, so the data model and the code that produces it never drift, and a backfill engine rebuilds source records faithfully enough to round-trip, so history is correct by construction.

Canonical model: a shipment is the unit of economics, and it touches a dozen parties. Each one is the same Organization record in a different role, not a separate copy, which removes a whole class of sync bugs. Cost and revenue are tracked separately with proper multi-currency, and consolidated invoices are reconstructed so nothing double-counts.

Secured REST API: clean, paginated, and filterable, locked down in layers, single sign-on, deny-by-default, role-based filtering, and a financial-data gate, with the most sensitive data protected at the database itself, not just the app.

Self-service reporting: staff build, group, pivot, chart, save, and share their own reports and dashboards with no spreadsheet exports. Every report renders off one typed result contract and respects the viewer's own access level.

AI access and CRM sync: an AI assistant queries the operational data through a read-only interface gated by the same access rules as everything else, and the CRM stays in sync with what's actually shipping.

DjangoNext.jsPostgreSQLCargoWiseAzure SSOMCP / AI

Power BI

Executive dashboard suite

15 dashboards from scratch, including full financial reporting

2mo → 2wk

build time

Challenge: Leadership flying blind. Financial reporting was manual, and every report request went through a single bottleneck.
Solution: Built 15 Power BI dashboards including full executive financial reporting. Analyzed 188,000 rate records to surface pricing insights.
Result: Eliminated manual report requests entirely. Leadership self-serves every metric, and the rate analysis changed the company's quoting strategy.

Click through one I built, it's live with synthetic numbers →

Power BIDAXPower QuerySQL ServerAtlas API