Hiring a data scientist - some thoughts from the other side
We recently recruited for a new analyst. As always, this made me think about how geek teams work, what our team needs next and how this whole ridiculous recruitment dance happens.
During this process I did a few google searches - just casting around (as always) for information and ideas about how to recruit for a kick arse data scientist. I was struck my the vast number of 'helpful' articles aimed at the job applicant and how few addressed how a recruiter can strive to do their job well.
And then as I was notifying unsuccessful candidates, I was struck (again) by how surprised (and even grateful) some candidates were that I bothered to contact them.
Frankly, this all irritates me. So here are some of the insights I have developed over the years. For recruiters. Because we are the team that needs to lift its socks up.
1. It's not about 'winning' the job
I've had a couple of people comment recently that females don't "sell" themselves as well as men in their applications. This often comes up in discussions about why there are not as many women at the top end of tech. There is a perceived idea that men are more willing to 'talk up' their achievements than women.
Right or wrong - that's not what I want to talk about today.
The nefarious (and subtle) part of this conversation is the unspoken assumption that women (or whomever) need to play up their CVs to 'compete'. This, in turn, comes from an assumption that its all about winning the job.
I think that its recruiters that need retraining here, not the applicants.
I think that this mentality lets the recruiter off scott-free. The job you are offering is not this great prize that everyone should compete for at all costs.
2. Recruiting is a conversation...
...A slightly awkard conversation, sure. But still a two-way exchange of positions to find the best fit for both.
Ultimately, we need someone to fulfil a role. Through the recruitment conversation we need to understand what bits of the role the person will slot into and which will need development. We want to tell how the team will need to respond - will other roles need some adjustment?
On the other side of the table is the job applicant. There is nothing better for me then when I can see your successes and your weaknesses clearly. If you are not suited to the role, it will make you miserable and you don't want it. Trust me.
So recruiters - don't be scared of honesty. Instead:
- Consider, argue, develop and know what you are looking for. Understand the role's responsibilities, the technical requirements, the communication skills, and the team dynamics.
- Weed out the ambiguous and the embellishers.
- Test the technical skills directly. If you are hiring people in a specific programming language, ask them to code in it. But look for more than pass/fail - are they across it? Are they rusty on the syntax but have good programming style? Who are they and what would they need to fit into the role?
Some organisations train recruitment teams in this but I've seen many in the tech space and in the startup space that really aren't clear on these points.
3. Suck it up recruiters - it's worse for them
I know from my own experiences applying for jobs (and watching friends go through the process) that it's often awful. You might be on standby for weeks without feedback. You may never hear another word. You sure as hell can't count on useful feedback.
I live by a couple of simple rules of communication with applicants:
- Try to get through it quickly. Sometimes there are genuine hold ups, but you have to recognise that people are hanging on this. Get it done. Prioritise it.
- Talk to them.
- We email all unsuccessful candidates and I always offer feedback.
- And I ring all unsuccessful interviewees and tell them (no email). Its really hard. I hate it. I think that many recruiters avoid it because it's hard to be the bad guy. But - it's harder to hear that you missed out. And it is so much worse if you don't get any information about why. So suck it up princess and talk to the human beings that have taken the time to test out if they belong on your team.
If you talk to someone, you are able to provide genuine feedback. Tell them if their CV let them down. Tell them if they were great but as a team you deliberated and you really needed X and Y, not Z.
I've had good conversations with people who wanted feedback. I've had people call back the next time I recruited and list how they had gone about developing the skills I told them they needed (yes, they got the job that time).
Have you got other tips or thoughts about recruiting for a data science team? Let us know in the comments below.
A slight disclaimer - These are ideas that work well for professional service/technology/geek teams. If you are hiring for different roles, or your company culture is that of the old-school corporate ladder, these tips may not work. But please read them anyway - start bringing down the system from within.