product work

I've worked in product roles at two orgs and 3 business units, for a total of 6 years of product experience. My experience thus far has been primarily in the realms of manufacturing, life sciences, and biotech, but I am firmly of the opinion that the lessons I've drawn from this experience are generalizable to any complex workflow.

For a standard two-page résumé, click below.

I've also (because I love writing!) written a more detailed version of the key points of my résumé... just for fun?

Sepasoft: streamlining manufacturing

Starting in 2018, I established the product management function at Sepasoft, where we delivered a suite of MES (Manufacturing Execution Systems) software solutions. In that role, I led the release of two new products – the Business Connector module, for automating and simplifying communication between the MES layer and the ERP (Enterprise Resource Planning, like SAP or Oracle NetSuite, etc), and the Batch and Procedure module, a standards-based tool for automating complex, interconnected manufacturing workflows.

My time at Sepasoft was key to my gaining an understanding of B2B software sales – finalizing sales where initial and ongoing expenditures measure in the many thousands of dollars, not to mention a massive effort investment, often involving third parties, requires the provider of the solution to establish a relationship of deep trust with prospective clients. Furthermore, the relationship between the provider and the user is extremely long-lasting and high-touch; as product manager, I was just as likely to be considering the needs of an existing user considering and expansion as I was pondering a new feature to support a large new contract.

Pandemic Response Lab: diagnostics at massive scale

In 2021, I departed Sepasoft and continued my product work at the Pandemic Response Lab, a business unit of Opentrons. PRL performed the lion's share of COVID-19 sequencing in New York City, processing many thousands of samples per day, including up to 50,000 diagnoses per day during the Delta and Omicron "surges." At PRL, I led the team working working on the "DaViD" LIMS (Laboratory Information Management System), which provided a sturdy framework for tracking samples as laboratory technologists received COVID sample tubes, sorted them into racks, transferring those tubes contents into individual wells of 384-well plates, and finally using qPCR to assign diagnoses to each sample. Our tooling delivered instructions and collected results from

Most notable about our solution was the tooling around evaluating the results of a qPCR "run" – from one unified interface, licensed PhD scientists could evaluate the PCR "curves" and understand the relevant virality of each sample, and assign diagnoses. By all user testimony, this was a unique solution in the space due to its seamless connection between device output and in-context understanding of the samples involved. Our tooling greatly simplified the cognition required to

During my time at PRL, along with generally pruning the workflow and using techniques like poke-yoke to avoid errors and data loss, we massively enhanced our users' understanding of the samples in flight by adding key data around how and where samples were stored, and the pedigree or ancestry of each sample plate; e.g., what four 96-well plates were transferred from to create this 384-well plate? In the words of the lab director, this meant that when it came time to find a "lost" sample needed for a re-run, that search became a 30-second process, instead of an hour- (or even hours-) long one.

We also oversaw the expansion of our footprint to locations in Rockville, MD and Orange County, CA. At PRL-OC in particular, our extreme focus was on very minimal turnaround times (as low as 6 hours) to support ongoing entertainment production in nearby Los Angeles. In addition to the software-side work of spinning up and customizing additional instances of DaViD, I also visited each lab and performed essential setup and training exercises – I even got to help build some robots!

At PRL, I was afforded the remarkable opportunity to be promoted to Senior Product Manager, and to be given a mentorship role to a growing team of product managers, three of whom had not yet worked in such a role. I met with each of them weekly, and worked to help them come to grips with their relevant areas of assigned work. This was an incredible experience for me, and I learned so much, not just about the work of leadership, but also from all the prior experience these incredible folks brought to the table.

In the latter days of PRL, we were tasked with massively expanding our LIMS to be test-agnostic – to allow for the testing of many pathogens. In this time, our software delivery was extremely progressive and fast-moving – within a matter of months, we had revamped the existing tooling to indicate the relevant test and diagnoses in extremely user-friendly fashion.

However, in late 2022, business leadership at Opentrons determined that the demand necessary to sustain a large-scale, automated diagnostic lab no longer existed, and closed all three locations, and laid off all staff – no fun! However, I managed to get retained, and move over to another Opentrons business unit, Neochromosome.

Neochromosome: simplifying next-gen sequencing

Neochromosome, again based out of New York City, is a biotech startup that performs key roles assisting medicine manufacturers and bio-ag companies with various genetic processes, like medicine discovery, construct scaleup, and other such efforts. At Neo, I've had the remarkable opportunity to work with an incredible set of PhD scientists and research associates.

The primary task assigned to our me and our close-knit software team has been to simplify and error-proof the process of performing genetic sequencing – almost every effort at Neochromosome, whether client contract or internal R&D efforts, not to mention those who pay us for this service in particular, culminates in having the relevant genetic data analyzed and "sequenced." In practice, this means delivering to the client, internal or external, a readout of the DNA or RNA that makes up a genetic construct.

In a similar fashion to PRL, my team worked to produce a LIMS system that would help them track samples throughout their entire workflow. Unlike at PRL, however, there was not an existing software workflow in place. We did, however, have an incredible platform, called "Rosalind," that we could build such a workflow upon.

Upon selecting this project for work, we delivered possible the key piece of the sequencing workflow puzzle – tooling for generating the so-called "sample sheet," or set of instructions for an Illumina brand sequencer, within a very brief period of a month or two. This deliverable replaced an intensely manual process, based primarily in Google sheets and home-grown Python, that was extremely cumbersome and error-prone – each sample sheet would take well north of an hour to produce and check for errors, not to mention re-dos if errors did occur. 50 hours a month of our PhD scientists time would be a conservative estimate of time saved. From there, we worked to account for more and more of the troublesome tracking efforts that populated the workflow in the sequencing lab.

Perhaps the crowning achievement of this effort was the "Evaluate+Release" functionality we spearheaded around collecting output data from the sequencing devices, pre-assigning pass/fail per-sample based on per-client requirements, and automating the process of delivering result/output files to our internal and external clients. All our PhD scientists had to do was open the relevant page in Rosalind, perform a quick validation (as little as a few seconds, in some cases), and hit the release button. This process replaced a process wherein our scientists would use legacy tooling to evaluate results, and then manually generate a list of samples to be released, that they would then hand-deliver to a software engineer for delivery to client. Again, we saw a savings frequently on the scale of 40+ hours a month of PhD-level effort due to these changes.

Additionally, we produced a client-facing web portal, codenamed "Hopper," for collecting sequencing orders. Previously, clients would email our scientists order forms, who then had to perform the work of introducing these samples into Rosalind manually. Currently, we are underway on efforts to handoff these orders to our CRM as well!

At the publish date of this article, we have built an all encompassing set of tooling that took Neo's sequencing lab from entirely manual and spreadsheet driven, to extremely automated and hardened.

other roles

Prior to working in product, I worked in QA and support roles, which really gave me an close view on the sorts of things that bother users and clients – the support and QA perspective is something I try to still hold close in mind as I plan roadmaps and major new features – retention and expansion is just as driven by feature quality as by novelty, and perhaps even more so.