(This is a special edition of Marct AI — the first of a two part series sharing insights from interviews with AI/ML/DS hiring managers across companies from seed-public on how they approach technical team building. Regularly scheduled job opportunities and my thoughts coming in a few weeks.)
Attracting mission-aligned technical talent is a mysterious art, let alone the added complexity of hiring a new species of ML engineers and data scientists—people that make conversational sales assistants, models that predict churn, and systems that recommend “best” products for your customers. If you’re a startup founder or early-employee tasked with growing or configuring your AI team, or think you might be one day, these tips are for you.
After hiring 15+ ML PMs for my AI incubator and matching dozens of individuals in related opportunities, I wanted to learn how to engineer an optimized hiring pipeline for AI/ML. So I interviewed experts at startups in various stages (seed to post-IPO), synthesized common themes, added some of my own takes, to produce this write-up towards creating a better and more transparent end-to-end process for hiring in machine intelligence.
I hope you’ll find the iterated learnings of recruiting and reflections to be an actionable sort of playbook for operating an AI-first company and challenge some of your assumptions about hiring. I’ll break this essay up into two digestible parts: part 1 on finding/locating ML talent, part 2 on assessing/selecting ML talent. I’d love to hear your thoughts, feedback, or if you choose to incorporate anything here! You can reach me at alex@marct.ai anytime:)
I’ll concentrate on how early-stage companies hire, and briefly touch on processes of Lyft, Syngenta and FinCo, because it’s helpful to see how later-stage, more resource-plenty companies function and work backwards in scaling.
The people I spoke to are:
1. Grace Boatman - Head of AI at Recurrency (seed)
2. David Stanley - Senior Machine Learning Engineer and Computational Neuroscientist at Neurable (pre-series B)
3. Serg Masís - Data Scientist at Syngenta (on-track to IPO in 2022) and Author of Interpretable Machine Learning with Python
4. Michael Yoshizawa - Data Science Manager at Lyft (post IPO)
5. Lead Data Scientist at Public Financial Services Company, which I will call ‘FinCo’ (requested to anonymize name and company)
Special thanks to the Bloomberg Beta family for helping me design this project and reviewing the early drafts; and to our speakers: many thanks for your time, thoughtfulness, and wonderful insights. You can find more essays like this in the future and other resources for ML talent, founders, and companies by subscribing to marctai.substack.com (I’m building a community for ML talent too, fill out this form to join!)
Community-led growth is the expansion pack to supercharge your talent-scouting efforts and inbound applications that you should invest in.
Growing the community is not only fun for extroverts on your team, but can also act as a triple-purpose lever to 1. garner early advocates therefore word of mouth marketing, 2. transfigure product roadmaps, and 3. consider candidates outside traditional profiles but have proved their value-add via their reception within the community.
This has worked powerfully at Neurable, where the team grew their Discord community to hundreds of members by hosting frequent Hackathons and AMAs. The qualified candidates are easy to spot: they are enthusiastic/energetic (about learning or sharing their neuroscience project), positive-sum, and quick to demonstrate their capabilities through casual interactions with other community members publicly.
As a ML startup, there’s an endless stream of possible topics for your community — discuss the latest reinforcement learning papers, bond over common pains of dev-ops and model version control, debate if DALI is better than GPT-3. You might need to talk amongst yourselves (your team) to spark initial conversation. The world is your oyster.
How to get started?
Building and growing the community is a non-trivial investment of your time, it’s no doubt a high-leverage task as a founder in order to maintain a consistent cadence on customer research, determine feature priorities, and/or potentially strike gold on talent. You can properly calibrate this system and the time you want your team to spend, without taking too much away from their main tasks, and encourage community-led-growth to be a responsibility and touchpoint for everyone in the organization.
If you are a YC company like Recurrency, you’re in luck with the Work At A Startup job-board. But generally if you are part of an accelerator or have credible investors, take advantage of free distribution via a centralized portfolio job-board to increase the number of in-bound candidates of all kinds.
For senior-level machine learning engineers, personalized outreach on LinkedIn and within network referrals are often necessary. Also ask: who or which co do you aspire to be culture-wise? For example, Recurrency targets companies such as Stripe and Vise, to locate engineers that possess both the “startup mindset” and similar values.
At larger startups like Lyft, Syngenta, and FinCo, job portals alongside incentivized employee referrals are sufficient to create a pool of applicants. Top of funnel are handled exclusively by internal HR or external recruiters. Data or AI team leads do almost no personal outreach. However, they do actively focus on two things:
Elaborating on 1. (leveraging colleges as a talent source), you don’t need fancy venues or high-budget events to start.
First develop a knack within the company of having all your early-employees being able to articulate the excitement behind the vision. Instead of vying for the attention of typical Ivy League or top institutions, take pride in discovering, convincing, and making the careers of students in under-rated universities. It’s easiest to start with your Alma Mater, since you’re oriented in the campus scene. Reach out to the clubs that you’ve been a part of previously or seem relevant e.g. AI projects club, entrepreneurial society.
Creating a progressive presence on multiple campuses is one sure-fire way of hiring faster and better from an exclusive and replenishing supply, compared to your competitors who may be still stuck with hiring from expensive job-boards with a questionable and unknown supply.
Once you have identified the right parties to engage with, offer to host a workshop—an event that is primarily educational and valuable for students, sprinkled in with a tasteful morsel of hiring messaging or advertising. Don’t underestimate the deceptive allure of good snacks, well-designed and actually wearable swag, and efficiencies of personalized outreach in closing the best candidates. If you do this correctly, great! Now you have an established presence, somewhere you can hopefully get first picks in talent, and sometimes in the best case free evangelists for your product (think Morning Brew Ambassador program in their go-to-market)
Interns often are hungrier because they need to prove themselves early in their career and set a high bar for energy. They can also be significantly de-risked hires, and if you’re able, they offer creative hiring opportunities (imagine the usual trial periods being extended to months).
On your path of building as a founder, you should be in permanent recruiting mode, and keep an eye out for emerging and passive candidates as a habit.
Until next time (more to come on talent assessment), let me know what you think!
Alex
