Services

Built from the whole problem outward.

Small and midsized businesses rarely have a tidy “website problem.” They have an order problem, an inventory problem, a catalog problem, an approval problem — or several systems that no longer agree. I design and build the software that connects those pieces into a working operation.

01 · Commerce & Marketplace Systems

Operating the channel, not merely advising it.

I built and led an Amazon Marketplace channel from its first listing to more than $30 million in annual marketplace sales, and have been responsible for the operation behind it for a decade. That experience covered nearly every part of marketplace operations: catalog architecture, feed automation, repricing, FBA logistics, suspensions, policy changes, and the thousand operational details between an initial listing and a mature revenue channel.

My deepest experience is on Amazon; I also have experience with Walmart Marketplace and the tooling that keeps multi-channel catalogs, inventory, and pricing coherent. And I’ve owned the content side of the channel too — more than a decade of product photography and listing video for marketplace catalogs.

02 · ERP & Connected Operations

Your business shouldn’t bend to fit your software.

Odoo is the rare ERP flexible enough to fit the business — if it’s implemented by someone who understands both the platform and the operations it’s modeling. I handle implementation, customization, module development, data migration, and the integrations that make ERP genuinely useful: storefront, marketplaces, instant search, inventory, purchasing, and accounting working as one system, so the team can stop living in spreadsheets and disconnected tools.

Odoo ERP at the center, connected to storefront, marketplaces, instant search, inventory, purchasing, and accounting. An AI layer for enrichment and digitization works alongside. storefront marketplaces search layer inventory purchasing accounting Odoo ERP AI layer
fig. 2 — one system of record, everything connected to it
03 · Custom Business Systems & Portals

Software shaped around how the business actually works.

Customer portals, administrative tools, order entry, ticketing and booking systems, inventory and workflow applications, product and content management — the systems that don’t come off a shelf because they have to match the operation. I’ve been building them since before many of those labels were familiar to the businesses using them, and the rule hasn’t changed: the software bends to the business, not the other way around.

Sometimes the honest answer is a hosted platform — Shopify or Wix — and I’ll say so, including when the cheaper option is the better option. When the business needs software no platform provides, I build it in PHP, Python, or JavaScript. That extends to mobile: when a business genuinely needs an app, I build it; when it doesn’t, I’ll save you the money and say so.

04 · Applied AI

AI put to work on specific business goals.

I use AI to build software — and increasingly, I build software that uses AI. Not chatbots bolted onto a website: working modules that solve defined operational problems. The problems I build for look like this: a catalog of thousands of products with half the data missing. Decades of printed catalog pages waiting to become structured, searchable content. A receiving dock where a camera reads an incoming package — dimensions, part numbers, product images — and AI matches it against the catalog.

The standard is the same one I apply to any system: does it achieve a specific operational goal, and can the result be trusted? AI earns its place in a module the same way it earns its place in my workflow — by being verified, not assumed.

An incoming package is read by camera — size, product image, part numbers — matched against the catalog by AI, and identified. size · image · part no. package arrives camera reads AI matches catalog item identified
fig. 3 — receiving, where inventory is identified
A printed catalog page is photographed or scanned, AI extracts the content, producing structured product data. printed page photo or scan AI reads & extracts structured data
fig. 4 — decades of printed pages, becoming data
A sparse product record goes to a web-research model; high confidence goes straight to the complete record; low or medium confidence goes to a second model; its high confidence also completes the record; medium or low goes to human review. high low / med med / low approved high sparse record model 1 · web research model 2 · deep research human verifies complete record
fig. 5 — enrichment: high confidence matches complete; humans review what’s left

Who this is for

I work best with established distributors, brands, and owner-led commerce businesses whose growth has outpaced the spreadsheets, disconnected platforms, and improvised tools holding the operation together. The common thread: consequential business-system problems, not miscellaneous technical tasks.