.png)
In the last few months, we’ve seen companies create products that change how we go about day to day exercises like writing and debugging through AI. As one of the hottest spaces, we partnered Paul Gomila, lead recruiter at Arena, to get a deep dive into how AI companies hire and how to stand out as a candidate looking to get into AI.
I think there’s some strong overlap with what is unique about hiring in AI and what is unique about us. To name a few things that stand out as important:
The pace the industry is moving at is pretty wild, I think there’s an emerging emphasis on applying new things that there aren’t a lot of guidelines for because of that.
At Arena, a lot of what we do is new across the technology we’re using, the problems we’re applying it to, and who we’re doing it for. We’re using some of the most powerful ML technologies out there (transformers, simulation, reinforcement learning, active learning) to build autonomous decision making systems for the biggest companies in the world. Almost everything we’re building is in production for an enterprise customer, so there’s an added layer of complexity to what we’re building. It can’t just work in a lab scenario or on paper - it has to work in real life. Our culture is oriented around this model of going to a customer and building something that provides immediate impact, and that informs everything we do on the hiring side as well.
There’s a generation of AI companies now that are doing really amazing things, but a lot of it is for individual users, or in the enterprise cases, it’s more about helping an enterprise do their own ML modeling. Those things are challenging in their own right, but the emphasis on application and forward deployment we have is a pretty unique challenge for candidates to jump into. Engineers and scientists at Arena have to straddle driving the technology forward while maintaining a focus on business impact, and do that quickly in low guidance situations. All of that boils down to needing really solid technical fundamentals and the ability to make sound decisions in novel contexts, so that’s what we look for, and what we evaluate for. It’s tricky enough that I had to re-learn how to recruit when I joined.
The culture we’ve built around ownership and autonomy is something we’re really trying to bring to life in our recruiting process - from org structure, interview process, job descriptions, titles, everything. I really admire how Adept does their job descriptions; they’re functional and contextual, it’s not a list of tools you’ll use or projects lined up for you to complete. I think broadly that’s a trend we’re seeing in AI - it all circles back to thriving in ambiguity and being able to own your problem space.
When we think about what success looks like for us, it comes back to the culture of ownership, it’s about being able to make decisions independently in novel contexts. Everyone is expected to be able to solve problems across the breadth of their role, so we have to trust everyone to do that. If you’re a Machine Learning Scientist here, it means you should be able to do everything from analyzing raw data to deploying the model yourself. The way we operate demands that we dig deeper and figure out who people really are, what they’re motivated to do, what they’re capable of, and why they’ve made certain career decisions. We care a lot about intrinsics.
We also care a lot more about technical fundamentals than specific expertise. Pratap, our CEO at Arena, mentioned in an interview with Primary Ventures, “We have a philosophy: when you think you need an expert, question that notion from every angle. You usually discover you don’t.” That has been proven over and over again for us, we almost tripled headcount last year, and a lot of those people have never worked in AI. A track record of having ownership and being effective when you do is way more important to us than expertise with a particular tool. Some other things we look for:
Green Flags
Focusing on any of the qualities above is a great place to start. More specifically:
Every interview process here is informed by the culture and the business. We need to know a lot about a candidate, so we want to design a process that is succinct, challenging (but fair), and can provide strong signals for technical fundamentals and culture add. Generally, we care a lot more about a candidate’s ability to problem solve across the breadth of their role than we care about their skill with any particular language, tool, etc, so a lot of our technical interviews are either designed as real-world scenarios they could encounter here, or otherwise focused on application of a skill instead of a tool.
Operationally it starts with a call with me, then some kind of take-home assessment (for technical roles, usually something custom-made in Coderpad), and then the final round typically consists of onsite technical interviews and a behavioral interview, back-to-back in one afternoon. This way, we can build momentum with the candidate and also make sure the whole interviewing team has their experience fresh in their mind for the debrief we do shortly after. We do our best to make a decision within 48 hours of that on-site final interview and follow up with offer details in this timeframe.
Pratap gives every new hire neon shoelaces when they’re introduced to the team at our all-hands meetings, we all have them. He used to do triathlons, so it’s kind of borrowing from the tradition of changing your laces to bright colors before the race, it’s celebratory. It’s like “We know you’ve worked hard to get here, now it’s time to keep going.” Also, every new hire adds a book to our office library and gets to say a bit about why they added it. I've always loved that.
We also hand out mission patches for different accomplishments, something Pratap was inspired to borrow from NASA. A good example is forward deployment to meet with customers or potential customers; this year we’ve had team members travel to Ecuador, Mexico, Germany, India, Canada, and various places around the U.S.
This article was published in partnership between Arena and Complete. You can learn more about Arena and their open roles by visiting their website.
