Come work with me

Two hires at Recast: Infrastructure engineer and data scientist

Hi! This is a departure from my normal posts. It’s an advertisement to come work with me, building Recast.

First of all — what is Recast? Recast is a way to measure how well your marketing spend translates into revenue or new customer acquisition. It’s one of the most important business questions imaginable, one that’s been around as long as there was advertising. The old quote “Half the money I spend on advertising is wasted; the trouble is I don't know which half” by John Wanamaker over a century ago is as true today as it ever was.

Don’t believe me? Believe Airbnb — there are countless other examples where marketers slashed or eliminated spend and found sales did not decrease. But yet, many businesses have been built successfully on paid media spend. So which is it? That’s what Recast helps answer.

We help answer it with a big statistical model that combines all the cool, cutting edge stuff: bayesian statistics, Stan, Jax, causal inference. It’s an incredibly tough nut to crack, but our model seems to do pretty well.

Also: We have clients! And real revenue. We’re modeling about $500 million in annual marketing spend for our clients. We have been bootstrapping the company but are so busy that we need some help, and we have seed financing to do it.

What help do we need? Two things. First — building our “job-running” infrastructure. Every time we run a model, we need to spin up about 30-40 EC2 instances, wait for them to complete, then run a second job to aggregate those results, and finally a third job to build the client-facing dashboard. We’re moving this to Airflow, and whoever we hire will need to know Airflow, AWS, Python, and be able to deal with the fact that a lot of the code that needs to run in these jobs is in R and Stan. See the job description here.

Second, we need a data scientist to help us work with our clients. Our model is fully Bayesian, which means that we need to smartly set priors on several inputs and intermediate quantities. We also often need to do analyses of the model results on an ad hoc basis, no small feat when it spits out something like 30k parameters. This person will need to know R and Stan and have a background in statistics and causal inference. See the job description here.

We’re 100% remote and open to international hires (no visa sponsorship, though). Ideally these roles will have a strong timezone overlap with US central time (or the willingness to work hours that do).

If you’re interested, or know someone who is, please reach out to tom@getrecast.com