Racer Biosciences

Most biology labs already rely on a DNA-assembly kit they trust. Rather than guessing from a lab’s general field, I built software that read the experimental Methods sections inside thousands of open-access papers and grant awards, verified the lab head’s email, and drove targeted sample outreach—feeding tester feedback straight back into R&D priorities and guiding academic buyers through university purchase orders and campus stockroom pitches.

A researcher wants clear experimental proof, while their university’s purchasing office usually requires an approved vendor account and a Net 30 purchase order rather than a credit card. I rebuilt Racer’s site as a custom storefront wired to Shopify and HubSpot to support both, and partnered with a teammate on a recurring LinkedIn science-story series backed by an automated drafting pipeline that respected her bench time.

After modeling showed that manufacturing in-house beat hiring a contract manufacturer, I helped secure lab space, built an auction scanner to buy used equipment, audited our ERP bill of materials line by line, wrote our first quality manual, and built the web app the team used to trace controlled documents, raw-material lots, and batch records through cold-chain shipping.

Alongside the reagent business, I worked with a teammate to build an internal beta of a DNA-assembly design tool using AI coding agents. Instead of assuming a researcher was ordering every piece of DNA from scratch, the tool started with what they wanted to build and the DNA they already had in their freezer—checking which pieces could be reused, which fragments needed to be ordered, and whether the sequence had structural issues that would make it hard to assemble. My teammate set the chemistry and manufacturability rules, while I turned the workflow into a working internal prototype.

With one teammate handling all of the laboratory work, our biggest constraint was her bench time: she was developing new products while also handling quality testing, customer support, and technical content. Six different R&D projects were competing for her schedule.

Rather than asking her to maintain a complicated financial spreadsheet, I built a straightforward project planner where she could enter the tasks, weeks, people, and supply costs for each idea. The model made the bottleneck unmistakable: the proposed projects required far more bench work than her schedule could support. I also pushed back on top-down market estimates, showing why a large industry market didn’t automatically translate into near-term sales for a small team.

I brought that same discipline into our regular management and board reviews—backing every slide with a workbook of primary sources, separating signed distributor agreements from partners actively placing orders, and keeping our hiring and R&D choices tied to the business we could actually support.