I recently completed the Data Science program at Galvanize. I did a lot of research and pondering about which path to take before finally deciding to attend Galvanize (the options I considered were: do it on my own through MOOCs, attend a University program or attend another bootcamp). So, after all is said and done, would I say it was worth it? Well, there was definitely some positives and I enjoyed going through the program, but overall I would have to say that the program does not deliver as much as it promises. But of course, this answer likely will be different for every person depending upon your background, current skills, interest level, and so on…

Here’s how I see the pro’s and con’s:

The Pro's:

• They have a very comprehensive curriculum, so you will be exposed to and have the opportunity to learn about a very broad set of DS-related topics. The capstone project will give you the opportunity to practice what you have learned, resulting in a "fun" experience with a nice add-on for your resume.
• The classes are held at their co-work facilities so you will find yourself in an atmosphere buzzing with tech start-ups, which instills a cool vibe and energy.
• You will meet and be surrounded by some very smart and friendly classmates. Through your joint trials and tribulations, you will become very close and develop strong bonds and a common sense of accomplishment.
• The instructors, in general, are very smart as well and know their stuff.
• One of the great side benefits is that you will receive some excellent help and tips pertaining to your job search. They will help you craft an awesome resume and cover letter, conduct mock interviews and point you to a variety of job search resources.
• As you know, it's real hard to maintain the discipline and focus to learn effectively from the multitude of free or low-cost online classes. So, in attending a boot camp you will be "forced" to learn and get it done in 12 weeks (after all, no one wants to see that kind of money go down the drain...)

The Con's:

• The downside to having only 12 weeks and covering such a broad amount is that they move very fast through the material so you will get the fire hose effect. Or in other words, the topics covered are a mile wide, but just an inch deep.
• Although the curriculum covers a broad amount, it is a canned "cookie cutter" program that has been more-or-less the same content and labs for quite some time. The instructors live with what they've got. They are very inflexible and unwilling to deviate from the standard agenda. It's their way or the highway.
• You only receive around 2 hours of instruction per day (roughly one hour in the morning and one in the afternoon). The rest of the time, you are either individually working (programming) on the morning lab or working with a classmate on the afternoon lab. So, for a large portion of the day, you're doing your own thing. The instructors tend to just sit at their own table and do their own thing. I found that I was learning more from my fellow classmates than from the instructors. This made it a rather frustrating.
• It's no secret that Galvanize has a high employee turn-over rate. Yes, they are growing very fast, and hardly anyone has been there longer than a year. In particular, they have a hard time keeping talented lead instructors around (they quickly get tempted and drawn away by higher paying and more exciting things to do). So, the classes end up being taught largely by former students who just finished the program weeks before you. True, some of them are very sharp folks...but it doesn't make you feel like you're really getting your money's worth.
• One of the main marketing messages that Galvanize likes to tout is their student placement rates (over 90%) and average starting salaries for their DS grads ($114K). With raised eyebrows, no one in our cohort remotely believed these figures. This has been borne out three months after graduating, with so far only less than half of our class having landed offers and of the handful that have most reported that their starting salary is ending up in the $80-90K range.

★★★★

Zipfian (as it was called when I enrolled) was a great experience for me. I emphasize that last part because the value of Zipfian largely depends upon what you bring into the program and what you hope to get out of it.

My background is in statistics, and I was coming directly out of a master's degree in that subject. I had also taken a number of courses in machine learning, so I was quite familiar with the mathematics used in Zipfian's curriculum.

So, using Tony's response as a style guide, here are my main points about Zipfian:

• You will become expert in the most relevant data science tools. Specifically, Pandas, NumPy, Scikit-Learn, and Matplotlib will become second nature to you. You'll also gain exposure to tons of others: NLTK, Mr. Job, BeautifulSoup, PyMC, D3.js, and at least a dozen others. Oh, and you'll become proficient in SQL and MongoDB.
  
  This was the most valuable part of the curriculum for me, and had I learned nothing else, it would have been enough. Dayenu.
  
• You will gain a basic mathematical understanding of classifiers and their search strategies. That is, you will have an intuition of what random forests, SVMs, and logistic regression are doing under the hood, and you will code several of those from scratch.
  
  Some students were disappointed with the level of rigor of the curriculum (while others found it daunting), but I wasn't. Mathematics isn't something that's learned in depth on a single pass; rather it's learned in increasing detail over many passes. What you will be getting from Zipfian is the first pass, roughly what you'd get from an advanced undergraduate class at a university. If you want to go deeper than the standard curriculum provides, you'll have plenty of opportunities to do so.
• There is (was) a lack of emphasis on statistical inference. In the areas where statistics and machine learning do not intersect, Zipfian emphasizes machine learning. At the time I thought this a prudent decision: how often, really, does one need a p-value? Now that I'm in industry I can tell you that inference – hypothesis testing, p-values, posterior densities – is actually quite useful. The curriculum could do more to teach these skills. And at the risk of being greedy, I would have loved to have seen a day or two on sampling algorithms.
  
  As Tony alluded to, there are a few new instructors who may have already addressed this.
• The community is really excellent. I can't emphasize this enough: you'll have a terrific group of instructors and classmates who are by your side, both during and after the program. If you're moving to the Bay Area and don't know a soul here, you'll be downright local by the time you graduate. Well, almost.
  
• The most important factor in Zipfian's quality is your attitude. That is, if you put your mind toward acquiring a set of applicable skills and don't give in to cynicism about job prospects or what other bootcamps are learning or things of that nature, you'll do well. If you're moody and indulgent of negative thoughts, less so. So if you do the program, give the instructors the benefit of the doubt that things will be OK, because they will be, if you're brave enough let them.
  
• Zipfian is the beginning of a career in data science, not the end of it. Will you have learned everything you'll need to be a great data scientist? Not by a long shot. But will you be in a stronger position to acquire and strengthen data science skills as your career necessitates? Absolutely.
  
  So should you do Zipfian? Only you can answer that.

I did not attend Galvanize, although I think the following review (pulled from this Course Report page), which matches what another recent Galvanize alum told me about the program, is a valuable viewpoint about the program as it stands in 2016.

The heavily advertised (self reported) stats are not reflective of the current course or hiring environment. In 2016, it will take you 3+ months to land a job coming out of the program and a sizeable portion (think 30-50%) of your classmates will not have any job after 6 months.

The program has been myopically focused on generating more revenue (students), and devotes the bare minimum of resources to graduates. The core product of the program (getting someone a job) was never well developed, but because of Macro trends (Hot Tech Market and Data Science Buzz) this went unattended. Now the tide has gone out and this program is swimming naked. Check your expectations … you won’t be hired by a prestigious company, have a strong brand or alumni community.

“Bad outcomes” as they are affectionately known, just don’t write posts like this – they still need the alleged connections and support of the program to get a job! The indifference the school has shown them has been shameful and prospective students deserve to know what they are getting into.

tldr: Galvanize is one of the better programs, it does a good job on the curriculum, but this model of education is still under development. Enroll in any Bootcamp at your own peril.

To management;

There is a lack of Leadership, Vision and Passion in SF, you can tell the CEO works in another state. Stop reading your own press releases about the future of education, and fix your product… Reduce # of students going through the program, & focus on outcomes

I will say — at least anecdotally, I have indeed heard that getting in contact with early cohort Galvanize alum is difficult/impossible, even though they have some of the best jobs of people who have graduated from the program.

Employers apparently don’t really add a premium for graduating from Galvanize — something to keep in mind if you think of it as a meaningful credential.

Still, it seems to be the best data science bootcamp out there. Galvanize hosts a lot of conferences, etc. at their SoMA space. Definitely take advantage of these network effects if you can!

★★★★★

Galvanize is simply the best data science bootcamp that exists. I'll walk through how I came to that decision, alongside sharing information and experience coming out of it. For you to get the most out of this program, you should be a rockstar with stats, math, and programming background and good work experience. As a background, I came in with a PhD in Materials Engineering (I mostly did computational image processing), and worked several years in R+D at Intel developing next generation computer chips.

I knew 1.5 years before joining Galvanize that I would transition from my role at Intel to data science. I applied to every bootcamp, and chatted with members from:

• Galvanize
• Insight (PhD only)
• Metis
• NYC Data Science Academy
• Data Incubator (PhD only)

Additionally, I read blogs from members who had gone through each program. Below was the most influential blog I read from Ike Okonkwo. Additionally, I have added my personal blog detailing the experience at Galvanize as well (more up to date with current curricula).

Zipfian Academy - All 12 weeks - Ike Okonkwo (Winter 2014 Cohort)

Blog - Scott Cronin (Fall 2015 Cohort #9)

I went to Galvanize because it was:

1. The most intense 12 week data science bootcamp
2. The most well designed and iterated curricula (now on cohort #12 as of March 2016)
3. Well rounded with additional focus on resume, recruiting, and interview prep
4. Has the single strongest Data Science alumni network. Literally there is someone from Galvanize at every tech company that you can connect with

Coming out of Galvanize, I obtained 5 job offers within a 2 month spread and had my choice of what I wanted to do. Note - you must be talented prior to coming into Galvanize. The interview process has 4 steps to verify you will be successful in the program as it is too intense if you do not have a background foundation in programming, math, and stats.

It is important to realize that YOU have to bring knowledge to the program so you can share with others. Likewise I learned a ton from others in my cohort. This is what makes the program great. Despite not having a CS background, I pair programmed extensively with those who did to become a strong coder. Additionally, once you graduate, you are connected to the single strongest data science community in the world. Two of my job offers came from the Galvanize community.

Hi,

I am from the very first Data Science immersive cohort at Galvanize-Austin. Before the program I used to be a Postdoc at UT-Austin and now I am a Data Scientist at Velexi.

My experience at Galvanize was above and beyond my expectations. The program covers a great deal of what you’ll need as a Data Scientist. The instructors are way better than in academia (I know that may not say much...). They are super smart (also in how to teach a topic) and they care about their students. The lead instructor brought some of his Data/Computer Science industry experience to the program which was a great plus. I am very pleased with the amount of experience I acquired with all the individual programming and pair programing Galvanize offers, and the environment helps so much mastering it (all the help we can get from each other and instructors from 9:30am to 6pm for 3 months). It was such great investment for my career!

Galvanize is having a hard time finding qualified enough people to join the program (specially the level of programming skills required to pass the entrance exams), that is why now they also offer a 8 weeks on just Python programming (twice a week at night), and the money invested on that can go towards the Data Science immersive after. Such a great idea to solve the need of have students starting with already a good level of programming to be able to survive the course and keep the high quality of the immersive.

Hi there! First of all, I'm super excited that you are applying for the Data Science Immersive (the DSI). I'm a DSI Instructor at Galvanize Austin, and I'd like to help steer you in the right direction.

The biggest piece of advice I can give you is this: It is in your best interest not to know the precise questions we ask during the admissions interviews. We strive to only admit students who will succeed in the industry as Data Scientists. We care deeply for our students, and it is for that reason that we only admit students who will succeed post-graduation.

But you should still study for the interviews! Here is what to study for each:

1. The takehome: Study SQL, basic hypothesis testing, and be sure you know python before you begin. If you need to learn python, try Google’s Python Class.

2. The python interview: This is just about testing your ability to write a program quickly (which is a vital skill you need in order to be successful in our immersive program). To prep for this, I'd suggest doing exercises on Codewars: Train your coding skills.

3. The stats/ml interview: I'd suggest taking Stanford’s free online Statistical Learning course to prepare for this interview.

The last bit of advice I'll give is this: It's okay if you don't know the answer during the interview. Just be honest and say, "Great question, I don't know, but I'd love to figure it out. Can you give me a hint?" A lot of what we look for in these interviews is the ability to learn quickly and on-the-fly. If you can solve the problem or answer the question after getting a hint, that's awesome!

I completed the Galvanize Data Science immersive boot camp, and afterward because of my background was asked to conduct technical interviews for the boot camp, which I did for a few months.

Putting it nicely, the answer is no— no one is rejected from the Galvanize Data Science immersive program. However, each person needs to pass the screening interviews before being allowed to enroll. And putting it not so nicely, not passing the interviews could be called “rejection”. But you can apply more than once.

The interviews examine your knowledge in relevant skills (Python coding and statistical/machine learning knowledge). The interviews aren’t just meant just to evaluate candidates, but also meant to teach candidates unfamiliar statistical learning concepts to see how quickly they can learn them and use them.

Galvanize has high standards for entry into its DSI program, not just because they have fewer spots than interested people, but because they want their boot camp graduates to be very well prepared to do data-scientific work after they complete the program. One way to make sure that people have success in learning the material covered in the program and, even more importantly, gaining employment in a data-scientific role is by making sure that people entering the program are as well-prepared as possible to learn the material and impress interviewers as to their fitness to do the work. So in a way, the screening interviews are a dress rehearsal for the even more intensive, more critical interviews graduates will face after graduating.

Most certainly, the application process is fairly difficult as far as non-accredited accelerated courses go. The first step eliminates the majority of people who are denied, which is a 3–4 question take home test that you have 4 hours to complete. Then there are 2 hour long skype interviews in Python and Stats to weed out those who just copied answers to the take home from the internet. Personally I made it past the first two stages only to be denied at the stats portion. They said just study up on stats and reapply, I have just passed the first two stages yesterday and have the stats interview tomorrow. I hope I don’t get denied twice as this time I already left my job.

First of all, congratulations on making the decision of joining the Data Science community! For full disclosure, I am a Senior Data Scientist at Metis in the Seattle campus. My suggestion is that you do due diligence on all the bootcamps you are considering. This should include:

• Visiting the campuses and getting a sense of the space,
• Meet the staff (this includes the careers advisor, admissions teams, instructors, program managers, and others),
• Request to talk with current students and alumns to get a sense of what they liked and did not like,
• Compare the admissions process, the career support, the background of the instructors,
• Ask about what percentage of alumns are employed (ask what qualifies as employed) and in what types of companies and roles they are working in,
• Compare the cost, duration, curriculum, hands on projects, etc.

There is no perfect bootcamp; however, there is the best choice for you, and only you can determine that.

I cannot speak about other bootcamps; however, at Metis we have a 12 week fully immersed program (9am-5pm). We have pair problems, lectures, guest speakers, tours of local companies, mock interviews, and work thru 5 projects in exploratory data analysis, regression, classification, natural language processing and unsupervised learning, and your final project is what we call the Passion project where you select something that interests you. We are accredited by ACCET which means that we have a 3rd party supervising our curriculum and makes us accountable for having 70% of our alumns employed in the field of data science.

Good luck on your research!

No bootcamp, degree, book or course will turn you into a hirable data scientist. They are all simply learning opportunities. You will turn yourself into a data scientist or you will not…that is the truth about everything in life.

No matter how you pursue it, it will be a ton of really hard work. If you are willing to do the work then eventually you will get there. If not, then no amount of time in a classroom will matter.

And the work continues after graduation on into your career. The reality about data science, when you get past the big salaries and hype is that there is no one path into the career, through it or easy way to be good. It is always challenging…always!

It's a bit awkward because after a week of studying purely Data Science(python and Java), I get tired and miss JS. But then when I switch to studying JS(engineering concepts) and study that purely for a week, I get a little tired and miss Data Science.

Ha! I know exactly where you're coming from here. Even during the time that I spent at Hack Reactor I frequently spent time looking at other topics and was (thankfully, in hindsight) reigned in by my technical mentors to focus on the tasks at hand. I think, to a degree, it's natural for most software engineers to want to be involved in everything.

I was sitting around with a few HR Grads in the Douglas (awesome alumni lounge) the other day and we were talking about other neat immersive programs that exist and most of us came to the same conclusion - we'd spend a period of time working then dive in to a new program. Because seriously, why not do both?

For that reason, I don't recommend one or the other; do the one that interests you most at the moment. Once you've graduated and spent time applying yourself and your skills (assuming you want to, I could see some just going from one to the next if their finances permitted) immerse yourself in the next challenge.

Initially I put a comment on Mike’s answer that I found to be deleted by him or Quora. He has a great answer with valid points that I agree however, I felt it was somewhat discouraging to newcomers in the field of data science and there was lot of emphasis on SQL. So I decided to put it as an answer below.

Like in any established profession, experience counts more than anything else and so in the field of DS. However, that does not mean that new grads are not groomed. Lots of big and small companies are focusing on it these days. How is that possible? Because, at least some experienced ones are taking the lead in training the next generation of data scientists. You can find positions similar to the below one quite easily (it is one of the most competitive position and demanding place though in data science). Experienced ones are not welcomed in this case.

DS / ML demands multifaceted skills. Besides obvious ML algorithms, you need some core CS or SW knowledge, statistics, domain or product knowledge and so on which is rather broad. Obviously, no one is going to master it in 2–3 years. It is also a rapidly evolving field. In fact, just because of that it may actually be easier for some new grads to get an opportunity in DS or ML. If you have cutting edge research experience in computer vision or NLP in grad school, I am sure you will have a better chance than many experienced ones.

You can not escape SQL if you are dealing with data. However, I found SQL to be the easiest to pick up. I know lot of people who did not know much SQL but were expert in another area when they entered in the field of DS and were able to pick up SQL pretty easily. The kind and amount of SQL you regularly use in DS is different from what you do in a DBA job. Also, if you already use Pandas for data manipulation, which is trickier than SQL, you can jump to SQL easily. I prefer to use SQL whenever it is possible which is fast and easily readable. In case you are not dealing with tabular data, such as if you are working in computer vision or NLP, your SQL usage may be very limited.

So does data science worth it? Absolutely.

This is the age of data revolution as the next phase of the internet revolution in the late 1990s. More and more of our health, business, personal and daily life decisions are becoming data driven. We are “online” more than before and generating lot of data that are useful. Whether you like it or not, now you have to understand data very well. To make it a career doing some “science” out of data? Sure, it is a new and exciting career with great demand. Definitely there are challenges ahead, like in every other profession. In fact, it won’t be wise if you don’t realize it.

I recommend you read Dr. Harari’s Homo Deus, especially the last chapter about data religion. I feel something like that is coming in near future.

I guess to make a fair comparison you’d really have to compare 3 year mphil with 3 month bootcamp + 2 .75 years of whatever follows.

If i were hiring you, I’d prefer these in order:

• bootcamp + nearly 3 years work experience
• 3 year mphil
• bootcamp only

So, the bootcamp + experience, in my opinion is the best. However, I’m not sure how you’re going to get that experience right after bootcamp. I certainly wouldn’t hire you, unless you were already an awesome coder.

You might get lucky and score a decent job after the bootcamp if you’re smart / interview well. Likely, you wont - and if it’s not a great job, you probably won’t come out if it much more qualified than when you went in.

Mphil on the other hand - you can probably get a decent job right after as a junior data scientist.

So, if you’re usually the smartest person in the room and feeling lucky, and if you’re already a great coder - I’d go for the bootcamp and try to impress my way into a good data engineer position and learn data science on the job.

If you’re not so much and want some more reassurance of a decent job, go for the degree.

There's 3 processes. The take home admissions test which comprises of an easy SQL that you can google. The python test consists of finding the lowest change return OR writing a simple text reading program which return word count, average word amount, number of sentences, etc...

The python interview consist of sorting an anagram. They want you to sort it in a dictionary because it's faster. They sometimes ask to return the mean/mode of an array or remove the max and min from an array and find the mean of that array.

The Stats interview comprises of conditional problems, sometimes binomial. There's a simple regression analysis, and a basic question on interpreting complexity of a test/training set. Which complexity you would chose and why, generally, stick to where the lowest MSE is possible. You'll also be asked to interpret an ROC graph. Finally, you need to know what false-positive, true-positive, false-negative, and false-positive in logistic regression. You're going to be shown one and asked to explain which section is which. Also, know what outliers are in a graph and.

It's pretty simple, but you can completely fail the stats interview as long as you have Data Science certificates from Coursera on your Linkedin. I don't suggest that you link your Linkedin when applying. That plays a major factor.