Wednesday, 10 June, 2020 UTC
Summary
When I was a candidate applying to Benchling, I had questions for my interview panel just like they did for me. I was interested in the life sciences, but it was a new industry to me. So I asked the panel about the industry, lab software, and Benchling in particular.
Now that I’m a full-timer here, I wanted to help answer similar questions for new candidates. Our recruiting team compiled the most frequently asked questions about Engineering at Benchling. I interviewed a few Benchlings across the company for their thoughts and summarized their answers here. In this post, we’ll cover:
- Do Benchling engineers contribute to open source?
- Do I need a biology background to be successful at Benchling?
- What are some integrations that are written with Benchling?
- Who are Benchling’s competitors?
Q1. Do Benchling engineers contribute to open source?
Yes, contributing to open source projects has been a great way to give back to an ecosystem that has benefited us all so much. Our larger open source contributions typically target the technical challenges we are facing. For example, we use the ORM SQLAlchemy and noticed slow inserts for thousands of rows, so Ray and Vineet wrote a Python module for faster batch inserts using SQLAlchemy (you can read more about it in the blog post we wrote).
I sat down with Alan, an engineer on the Molecular Biology team, to chat about his learnings from contributing to open source. I asked him how he chooses an open source project to get involved in:
When you take a look at the tools you’re using, there’s usually room for improvements. Sometimes it’s by improving performance or sometimes it’s by extending functionality. And it always gives you a better sense of how they work under the hood.
In 2016, Alan discovered Decaffeinate, an open source tool for converting CoffeeScript to JavaScript, when Benchling was faced with the challenge of transforming 200k lines of CoffeeScript code to JavaScript. At first, Alan filed some bugs and improved the error reporting. As he became more and more familiar with the code, he began to contribute more regularly and took over as the primary maintainer, filling in many missing cases, extending test coverage significantly, and fixing hundreds of bugs to get the tool to a point where it could be trusted for production use. You can read more about how Alan improved Decaffeinate to fit Benchling’s needs in the blog post he wrote.
Alan is also the author of Sucrase, an alternative to Babel that allows super-fast development builds. He noticed that Babel tries to compile a large range of JavaScript features to work with old browsers, like Internet Explorer. He built Sucrase when he realized that he could drastically improve upon Babel’s performance by focusing on a smaller scope. Sucrase assumes that you’re developing with a recent browser or recent Node.js version, allowing it to have less extensible, but more performant architecture compared to Babel. He introduced his project to other engineers at Benchling and, soon enough, everyone was using Sucrase to compile their developer environments!
In addition to larger open source contributions like Decaffeinate and Sucrase, our engineers also regularly make smaller improvements to tools that we use, like Postgres CLI, AWS + Okta, Apollo GraphQL CLI, Electron, Datadog, Sentry, and others. We aim to help document these tools too. Vineet wrote a blog post on how we automated database migrations in Postgres, which Postgres tweeted. And Saif wrote a blog post on front-end performance engineering with React, which was referenced by the official React documentation and had its code forked by the NFL!
Links
- Decaffeinate and From 200K lines of CoffeeScript to zero: making decaffeinate super-stable
- Sucrase
- benchling/sqlalchemy_batch_inserts, sqlalchemy_batch_inserts: a module for when you’re inserting thousands of rows and it’s slow
- Move fast and migrate things: how we automated migrations in Postgres
- Performance engineering with React and A deep dive into React performance debugging
Q2. Do I need a biology background to be successful at Benchling?
The short answer is no. Of course, Benchling’s product is specialized to the life sciences, so having a basic knowledge of biology is useful to understand how the product is used. But most of our engineers only have computer science backgrounds and are able to be very successful at Benchling. We provide ample opportunities for engineers to learn about life science R&D — including courses and science presentations. And many other functions at Benchling, like Product, Customer Experience, Marketing, and Sales, do have backgrounds in the life sciences. As an engineer, you’ll get to work with them, and learn from them, on a daily basis!
In addition to learning organically, all new employees complete an onboarding certification program when they first join the company. Through this program, you’ll learn about Benchling product, industry and science fundamentals. This program also includes a Biology 101 crash course that covers basic biology (like the role of DNA), established experimental methods (PCR method to copy DNA), and recent scientific developments (CAR T cell therapy and CRISPR genome editing). You will be able to complete your training within the product itself, using the same features scientists do, like Notebook entries and Molecular Biology tools. This way, new employees get a feel for what the product is like by dogfooding it. And as our organization continues to grow larger, we’re always investing in the onboarding experience.
Our onboarding module for industry fundamentals — dogfooding the Benchling NotebookBeyond onboarding, there are a number of ongoing efforts to broaden scientific knowledge throughout the company. For example, we organize onsite visits for engineers to see how our customers actually use the product. Usually this involves traveling to a customer’s lab and learning how they use Benchling to optimize their scientific workflows. We also host a Science Monthly talk series internally, where speakers with life science backgrounds give presentations on an interesting science topic. In the past, speakers have presented on the drug development lifecycle, biotech devices they’ve invented, as well as their Ph.D. research.
All of this is to say that a biology background is certainly not required to work at Benchling, but a willingness to learn about the life sciences is a huge plus!
Q3. What are some integrations that are written with Benchling?
Our customers use many lab instruments in their day-to-day scientific research. These include:
- Instruments that record temperature every minute
- Liquid handling robots that simultaneously pour hundreds of lab samples from one test tube to another
- Automated freezers that store and retrieve thousands of samples at exactly the right temperature
Benchling is a data platform for this lab ecosystem. For example, it can store the temperature measurements, concentrations and volumes of each vial, and physical location of each lab sample. For some use cases like measurements, the lab instrument is the input to Benchling. For others like liquid handling, it is the output — scientists describe the operations they need done, like “transfer 5 mL of the liquid barcoded BIO61 into TUBE29”, and Benchling sends those instructions over to the robot.
Integrations connect Benchling and other systems together. These systems can be lab instruments, but there are others too, like an online genomic database that the customer needs synced into Benchling. Benchling supports some of these integrations through out-of-the-box solutions like Benchling for Lab Automation, a plug-and-play interface between Benchling and lab instruments. We recently announced our partnership with robotics companies like Hamilton and Tecan!
— @jeremy_armetta
In addition to supporting out-of-the-box integrations, Benchling also provides a set of tools for developers to build custom integrations. Our Developer Platform team works to ensure that our platform is architected to be extensible. In addition to a REST API, they are currently working to support event-driven integrations (implemented on top of AWS SQS, Lambda, and EventBridge). That way, customers can chain together integrations that react to Benchling events. For example, populating new experimental data in Benchling with one integration can kick off another that validates that data or detects anomalies.
Architecture of an integration that connects Benchling to lab instruments and liquid handling robotsFinally, for use cases that are particularly unique or complex, our Customer Engineering team works with scientists to discover which part of their workflows can be best automated and then write the custom integrations to do so. Their past projects include:
- Automatically sync Benchling to a liquid handling robot’s actions
- Periodically export customer’s business-critical data to their data lake
- Poll automated freezers and sync physical sample locations into Benchling
For more details, the team’s Engineering Manager Neena and Product Manager Reed wrote the article Software as a Biotechnology: How Benchling Extensions and Integrations Drive Life Sciences R&D in our Benchtalk journal.
Q4. Who are Benchling’s competitors?
Benchling is a highly tailored data management solution for the life sciences industry. We operate in a fragmented ecosystem where there isn’t just one singular dominant competitor; rather, our competition comprises a range of different generic tools, point solutions, and other unified platforms like us.
- Generic tools refer to the broad category of unstructured tools scientists use to manage their experiment data. Many scientists still use pen-and-paper lab notebooks, Excel, or Google Sheets to capture their scientific data. These tools are competitors because they’re often the status quo. Scientists are accustomed to using them, even if it means their data are not well-structured, searchable, or easily retrievable.
- Point solutions are commercially available software that address very specific use cases. For instance, there are software tools that focus on providing an Electronic Lab Notebook (ELN) solution. While point solutions are optimized to handle specific use cases extremely well, they typically only support a small portion of a scientist’s entire workflow. This creates data silos and other inefficiencies when scientists are using an entire arsenal of point solutions and their data is spread out across different tools.
- Finally, unified platforms, like Benchling, offer a broad range of functionality and aim to store all of the data in one system. Most of these platforms, however, tend to be legacy software that haven’t kept up with the latest in both software and science.
Benchling operates in an industry that, until recently, has been relatively shielded from technological disruption. Our mission is to accelerate the pace of life sciences R&D by providing scientists with the most relevant and cutting-edge tools in the industry.
That wraps up this post of frequently asked questions about Benchling Engineering. Thank you to Alan, Alyss, Andrew, Isaac, Kayne, Prem, and Saira for participating in interviews to help answer these questions. Also, thank you to Amit and Somak for helping edit and review drafts.
If you have any additional questions, post them in the comments below!
Careers at Benchling
Discuss on Hacker News
FAQs about Engineering at Benchling was originally published in Benchling Engineering on Medium, where people are continuing the conversation by highlighting and responding to this story.