Florencio (Flo) Mazzoldi is the Head of Digital Technology at Ginkgo Bioworks, where he leads the development of the digital technologies that power Ginkgo’s foundries and Codebase. We are excited to dig into Florencio’s role and hear more about Ginkgo’s digital capabilities in shaping the future of synthetic biology. 

What is Ginkgo Bioworks and what does Ginkgo do? 

Ginkgo’s mission is to make biology easier to engineer. What we have created is a horizontal platform that enables cell programming. It’s a platform that we can apply to any biological product or industry.

What was the state of the platform when you joined Ginkgo? What is the future vision for the platform? 

With respect to the scale of the platform, in 2018, we used to do about 3000 strain tests every day. Strain tests are basically any measurement that we do on a strain that we’re building. Today, that number is more than 200,000. The vision of the platform is to continuously scale in a way that we can address more and more problems in the world. We also want to make it so that biology is so easy to engineer that we could address additional problems that are currently not even reachable. 

What were some of the core components that you really needed to enhance or harden in order to reach scale? 

You need a lot of automation and you need to standardize some of the operations. We have a Foundry model at Ginkgo, where there is a set of program teams that work with partners to set the strategy to deliver organisms. Then we have the Foundry team that runs high throughput biology. The Foundry team delivers services to the program teams in a similar way that you could think of a software company creating micro services that then can be composed to deliver functionality. The only problem is we don’t know how the cell works. So, we’re like programming computers, where you don’t know how the computer is going to react. 

We scale the amount of biology that we do each year 

That means a lot of pressure on your digital systems to be able to scale every year. The number one thing that we did in the last few years is we moved all our infrastructure to the cloud, so we can actually scale without having to buy so many servers. We also put an API on top of the Foundry. By creating the API, it allowed the Foundry to expose services to the program teams and the program teams to use the services of the Foundry, giving us a lot of visibility into operations. It also enables program flexibility without increasing the flexibility of the Foundry itself. 

We have also worked quite a bit on collecting and surfacing our Codebase 

This is another parallel to the computer industry. We get Codebase from all the learnings we have from running projects and try to reuse it wherever we can. An example of Codebase could be genetic parts that we know we’ve tried before and work well in certain conditions. We then surface this Codebase in our design tools, so that folks don’t have to go hunt for those parts. This can reduce the amount of things that folks try in the lab, which is great, because then it leaves room for other projects to be run in the lab without necessarily creating more machines or using more robots. 

How have you decided when to build versus partner versus buy?

If it exists, we try to buy and integrate as long as it’s not a core competency of Ginkgo. In the past, we’ve built, bought, and partnered with companies to build tech that doesn’t exist. If we see a need for a particular type of technology, and we see that somebody’s building it, and there is potential in that build-out, we will partner with the company to build it faster and also be able to use it in our use cases. The biggest barrier to buying, particularly on the science side, has been this extreme focus on scalability that Ginkgo has. We need to be sure that the tools that we’re buying are going to be scaling in a way that will enable Ginkgo to scale. We don’t want to be in a position where we cannot scale because some of the tools we purchased won’t scale with us.

So with the scale in mind and not willing to compromise on scalability, as you continue building out the roadmap and the platform, what are some of the biggest challenges?

The biggest challenge we’ve had traditionally is this trade off of scalability and flexibility. Because as I said, it’s a horizontal platform, we want to be able to basically run any cell program into our platform for any industry or any purpose of the cell. That means a very, very high diversity in the kinds of science that our teams need to do. So all our tools need to be extremely flexible. So with flexibility, you bring complexity. And with complexity, normally scalability suffers. 

Given the level of investment, both capital and time that your team has spent building this platform, does it make sense for every synbio company to follow in your footsteps?

As far as I know, we are one of the only horizontal platforms on synbio. So our platform requires a tremendous amount of flexibility. I’ve seen many companies taking a different route and going directly for a product type company. When you’re going for a product type company, your flexibility doesn’t need to be that wide. You don’t necessarily need to build such a platform at scale. However, with what we’re working towards, no one has to build these horizontal platforms. What we’re trying to do at Ginkgo is to make sure that whenever we start a project, we start with all the learnings from previous projects. Therefore, every project requires a lot less work than the project that happened previously. Hopefully in the future, it won’t make sense for folks to create their own platform.  

What are some emerging technologies that you are evaluating and/or excited about incorporating into Ginkgo’s Ecosystem?

The biggest thing in the last few months has been AI. I can tell you that, we’ve been looking at some of the advances on some of the public models, particularly large language models, that are on the protein side and less on the genetic side. 

Are you able to share any specifics or details around how you are applying AI or experimenting with AI? As it relates to protein design?

What we believe is that because we’ve captured so much data throughout these last few years, we should be able to take some of those models and retrain them in a way that we know will work in our settings. That’s the kind of work that we’re doing. We’re taking a look at some of the models out there and applying our data on them, which is one of our core competencies.

What are the missing needs for automation that you’re hoping for advances in? 

Standards, please, in particular on the data side. One of our main problems is that we have every instrument under the sun. So, being able to capture data in a standardized format so that we can compare across experiments or across runs of the same strain, or even compare context, has been very hard. Every manufacturer puts their own data spin on what happens. By advancing data standardization, it would make our analysis more streamlined. 

Is there a customer story that you can share with us? How are they working with Ginkgo and using your platform?

What happens is we engage with the customers that need to program a cell to solve a problem. I think that nitrogen fixation in plants is one that we’ve been focused on here in our agriculture vertical for a few months. We talked to a customer that needed a product that can spray on seeds that will help the seed do nitrogen fixation and we basically use the platform to build and deliver those organisms. This is a program that might take some time, however the impact is huge. Reducing fertilizer, increasing sustainability and additional yield on crops that reduces hunger.  Another big one that we work on is on cannabinoids, particularly the medical kind of cannabinoids. Instead of delivering them through the plant, be able to deliver them through a microorganism and brew it in a fermenter somewhere. We understand what the organism needs to do and needs to produce and we spend time with the customer building that in our platform. 

Thanks to Flo for taking the time to talk about the digital technologies that power Ginkgo Bioworks. If you would like to learn more about Ginkgo Bioworks’ Enzyme Discovery and Engineering Services; watch Emily Wrenbeck’s talk from Ginkgo Ferment 2023 here. If you would like to work with Ginkgo Bioworks on any future projects or collaborations, please contact the team with this link.