domingo, 23 de agosto de 2015

Goodbye GSoC.

Hello all,

This is my final entry for the GSoC. It's been one hell of a ride, nothing short of life changing.

When I started looking for a project, the idea was that even if I didn't get selected to participate, I was going to be able to make my first contribution to OpenSource, which in itself was enough motivation to try it.

I found several interesting projects and decided to apply to one of them, the one that I considered to better suit my situation at that moment. Before I got selected I had already interacted with a few members of the community and made a couple of contributions. I was hooked on OpenSource so there was no looking back. By the time I got selected, the GSoC had already met my expectations.

I found a healthy community in SciPy and I could not have asked for a better mentor than Ralf (Gommers). The community members were always involved and supportive while Ralf provided me with enough guidance to understand new concepts in a simple way (I'm no statistician) but not so much that I would be overwhelmed by the information and I still had room to learn by myself, which is an essential part of my learning process (that's where I find the motivation to do the little things).

After starting the GSoC I received the news that I was denied the scholarship to attend Sheffield and my plans for a master's degree were almost derailed. I then got an offer to study at the University of Toronto and this is where it got interesting (spoiler alert: I am writing this blog entry from downtown Toronto).

I went through the scholarship process again and got selected. I also went through the process of selecting my courses at the UofT. With Ralf's guidance and after some research I decided to take courses on Machine Learning, Natural Language Processing and other related topics. 

I can now say with pride that I am the newest member of the SciPy community which will help me in my journey towards becoming a Machine Learning expert or maybe a Data Scientist, that remains to be seen, but we already have some plans on how I can keep contributing to SciPy and getting acquainted with the pandas and Numpy communities. I'd like to see what comes from there.

As you can see, I got a lot more than I had expected from this experience, which I attribute to having approached it with the idea of searching for a passion to turn into a career. Naturally I found it, so now it's time to switch gears.

I would like to use the last paragraph of this rant to give out some thanks. Thanks to Ralf for walking me along to find my own path within the beautiful world of OpenSource and Scientific Computing. Thanks to the SciPy community, especially to Josef Perktold and Evgeni Burovski for providing so much valuable feedback to my PRs. Thanks to Google for organising an event like this, helping people like me with the excuse they need to finally join OpenSource and stop leaving it for later. And of course, thanks to the people in my life that provide me with a reason to wake up and try to be a little better than the day before: My girlfriend, Hélène, who keeps my head above the water when I feel like I forgot how to swim by myself and my parents, whose love and support seem to have no end. You make me feel like I owe it to the world to be the best I can be (or try, at the very least).

jueves, 6 de agosto de 2015

Progress Report

Hello all,

The GSoC is almost over. It's been a great experience so far and if you've been following my blog you know that I have decided to continue my involvement with the community, so this is only getting started.

With that in mind and some support from my mentor (Ralf Gommers), some tasks have taken a backseat while others have gone beyond the original intended scope. Most notoriously, the NaN policy which started out as a side note to a simple issue and has become the single largest effort in the project, not just in lines of code but also in community involvement (you can follow the discussion here or the PR here).

NaN policy is now in bike-shedding phase (reaching consensus on keyword and option names) but it is only the start of a long term effort that is likely to span for months (maybe years, depending on pandas and Numpy).

The NIST test cases for one way analysis of variance (ANOVA) are also coming along nicely and once they are done I will continue with the NIST test cases for linear regression.

Right now there are no major roadblocks but it is worth mentioning that Ralf and I have agreed to move the pencils down date to Aug 18th. This is due to the fact that I have to move to Canada soon to begin my master's degree, and this way I can travel to Toronto on Aug 20th to look for a place to live and also spend some quality time on vacation with my girlfriend Hélène, who has been a great support for me during this transition in my life. I feel like she has earned it just as much as I have.

Classes begin on Sept 16th. Once I feel like I'm settled into the new rhythm, I will get back to work picking up on loose ends or side tasks (like extending NaN policy's coverage) so the project will not suffer. I would also seek Ralf's guidance to start integrating myself into the pandas, numpy and possibly scikit-learn communities because I plan to steer my career towards data science, machine learning and that sort of stuff.

I will need to figure out where my motivation takes me, but this is a challenge that makes me feel excited about the future. GSoC may be almost done, but for me this is only just beginning and I could not be happier. As always, thank you for taking the time to read about my life.

Until next time,
Abraham.