Fellowship.AI Reviews

As I am currently on the Fellowship program (Sep - Dec 2019), I am able to share my experience from within Fellowship.AI. What I am about to share is not only my personal view but resonates with quite a few fellows who I was able to talk to via the Slack (which is the communication platform used by Fellowship.AI). If you decide to apply for the Fellowship program be aware that you are not going to learn anything useful about data science, machine learning or deep learning. There will not be any mentors who you will be able to learn from. You will not get any support from Fellowship.AI on your path to becomming a data scientist. You will get the opportunity to waste your time for a period of 4 months. You will have to devote the majority of your time to donwloading images from Google Images which Fellowship.AI is going to use for their own commercial purposes. Even more time you will have to devote to cleaning the image datatset. You are not going to apply any state-of-the-art deep learning methods. The best what you can hope for is training a resnet34 architecture using the fastai library (which provides a ready-to-use implementation of the resenet34 architecture) on the images you have downloaded. At the very start of the program there was a presentation by Arshak (CEO of Fellowship.AI) to fellows in which he talked about an exclusive machine learning community, about how much we are going to learn about machine learning and that we should be very grateful for having been accepted to the program. He advised us to think about how we could give back to Fellowship.AI. To me it was clear after the first month that there is no machine learning community. The only thing that is there is the commercial platform Platform.AI which exploites the fellows as free labour.

From the website, this is what they offer ( In brackets is my opinion) a) build scalable machine learning models with agile software development methodology. (Well, if for ' build scalable ML models ' they refer to download a lot of images and train a resnet, yes it is's true. This is what we do the 80% of the time ) b)They say: mentoring by experienced ML practitioners.(Not true. There is not mentoring at all.) c)pair program with other fellows (Yes, we are in the same location) d) apply latest research in deep learning, reinforcement learning, generative adversarial networks ( Not true, we are just applying a Resnet architecture) e) full-time for four months ( Yes, it is true) f) fully remote with optional in-person collaboration ( It is not fully remote, they asking the fellow to move specific locations) In addition, there are not ' real projects ' or ' real clients ' at all. ( I was based in Europe, but not in Paris )