Steve J Brown     About     Blog

Insight Data Science

Introduction

A couple years ago I participated in the Insight Data Science fellowship in Silicon Valley. Insight is a 7 week program to bridge the gap between academia and data science in industry. It has grown to offer fellowships in data science, data engineering, artificial intelligence, and more in multiple cities throughout the US and Canada. When I was deciding whether to do the program, I read several very helpful blog posts by past fellows including a three-part series by Ethan Rosenthal. This is my attempt to pay it forward and answer questions you may have.

What do you actually do at Insight

You work. A lot. The core of the program is a demo project that you present to partner companies demonstrating your skill in data science. For roughly three weeks you pour every spare moment into building, evaluating, and polishing that project. During those first few weeks, companies will also be coming in and describing what it’s like to work there as a data scientist. There will also be Q&A sessions and workshops by alumni and luminaries in the field. For example, in my session we had Q&As with Monica Rogati, Joe Gallagher, and Jonathan Hsu among others. During the remaining weeks, you present your project to hiring companies, prep for interviews with your fellow fellows, and do mock interviews with alumni. As soon as the program ends, you interview with companies that liked your demo and eventually land a job in data science.

Importantly, Insight assumes you start the program with 90%+ of the skills needed for a data science role. You fill in the remaining gaps by working on your project, learning from peers, learning from workshops, and preparing with mock interviews. It does not have a structured curriculum like a standard college course. There is a ton of support, but be prepared to drive your own learning and career.

My experience

I had a great experience at Insight. I created a website called git triage that would label Github issues and I was able to present my work at Github itself! After demoing at several partner companies, I got multiple call-backs to interview, and ended up having 3 offers by the end of the process – and they were all companies I would love to work for. I ended up deciding to go to Intuit as a Senior Data Scientist. Those are the results, but I also enjoyed the process.

During the program, I was exposed to the breadth and nuance of different data science roles. There are so many cool things people are doing with data! This exposure helped me refine the specific parts of data science I wanted to pursue. On top of that, the whole time I was surrounded by amazing, interesting, smart people who I am now happy to call friends and hang out with outside of work (and some at work too!).

Should you do Insight

I should start by saying, I think Insight is a great program! It worked well for me and, later, my partner. You don’t need to do Insight to get a job in Data Science, but it has several benefits.

One of the biggest benefits of Insight is getting your foot in the door. Cold applying to jobs sucks. The success rate of making it to the next stage is low and you often don’t even get a response. With Insight, the demo generally takes the place of a form application and has a WAY higher success rate. You are showing the hiring manager you can actually do the work instead of trying to hit the right keywords for a recruiter.

Another big benefit is being part of the Insight network. At this point, Insight has thousands of alums in over a hundred companies working with data. Chances are, if there is some aspect of data science you would like to learn more about (e.g. time series, productionalizing models, causal inference, leading a team), an alum has experience with it and is willing to share advice with you. Additionally, Insight fellows tend to be helpful, interesting, and fun to be around. I always enjoy meeting new fellows and catching up with old acquaintances.

Finally one underlooked benefit that I really enjoyed was seeing all the different ways companies are using data science. In a given session, maybe 30 different companies will visit. This is a great sampling of the variety of data science companies, projects, and cultures. I felt much more confident at the end of the program in the type of data science role I wanted to pursue.

That being said, it’s possible can get a job without Insight and there are some instances where you are probably better off going it alone:

  • If you already have stellar credentials, a good network in Data Science, and know what you want to do, just go get a job!
  • If you need a job right away and can’t afford to not have one for three months, Insight probably isn’t a good choice for you. Insight is free for fellows during the program (see below for a discussion of recent post-program income sharing changes) and additionally offers need-based scholarships to cover living expenses. But depending on your situation this could still be a stretch especially in a high cost of living area like San Francisco or New York City.
  • If you don’t know how to program yet or are looking for a structured curriculum to teach you data science, Insight isn’t a good choice. Insight assumes you are 90-95% of the way their in terms of skills and that you’ll pick up the rest through your demo project, working with fellow fellows, and some workshops. If you are earlier on in your journey, I would recommend online courses and side projects to build up your skills first.

If you are a parent, have a disability, or come from an underrepresented background, Insight is still for you! It is a lot of work, but it is an extremely supportive program that wants people from all backgrounds to succeed in data science. I have personally seen many people from all of these groups rock the program and end up with great jobs they enjoy.

Recent cost changes

Starting in Summer 2020, fellows will enter an income sharing agreement for 7% of their gross income for 2 years, as long as they make at least $100k/yr and get a job within 6 months of finishing the program. Previously Insight was completely free for fellows. Insight is hoping to use the extra revenue to make the program more personalized, achieve better outcomes for fellows, and support fellows as they continue in their careers.

With this change, some people have asked me whether Insight is still worth it. For all the reasons above, I still think Insight is a great program for fellows. Additionally, because you are interviewing at multiple companies at the same time, you can potentially command higher offers and spend less time interviewing – Insight claims their data supports this.

How to prepare

Preparing for Insight is pretty similar to preparing for data science jobs in general, but for those who may be new to the field it can be difficult to know where to start and what to focus on. Here is a quickstart guide you can follow:

  1. Learn Python. Some places use R and still others use Scala+Spark, but Python is the standard. I liked the course and book Introduction to Computer Science and Programming Using Python for learning Python with some CS fundamentals thrown in.

  2. Do a data science side project. Do an analysis or put together a machine learning algorithm on a topic you find interesting. Then practice explaining why you did the project and what you did. This is really helpful when applying to Insight and I think shows you can walk-the-walk when applying to jobs as well.

  3. Take Andrew Ng’s machine learning course on Coursera. This course does an amazing job of covering machine learning basics and is the gold standard. Although it uses the Octave language for the assignments, the community has put together python versions of the assignments that you can go through instead.

  4. Learn SQL shortly before applying. SQL is a necessity for almost every data science job to a lesser or greater extent. It’s fairly simple, but is easy to forget if you’re not using it so I’d focus on studying it shortly before interviewing. You can get into Insight without this, but you will definitely need to know it before the end of Insight. I like SQLZoo for this.

  5. If you are accepted to the program, I highly recommend starting to think of project ideas. Make a long list. This will make it much easier to hit the ground running once you get to Insight. Do not get overly tied to one idea. Many good projects are not good Insight demo projects for one reason or another and you will get a lot of feedback at Insight to help you choose the best project to show off your skills.