2022 was a long year. Tech market started to cool down and even some big techs started to layoff people, some friends got caught in the middle of this but were able to find a new place to work. Brazilian politics had a huge battle for democracy (that we fortunately win). I learn a lot of challenging things on the AI field.
In 2021 my focus on AI study was on data annotation and labelling. In 2022 I helped to customize and extend AI products focused on computer vision and object detection. This kind of projects requires a lot of efforts in the pre and pos processing, and a lot of additional steps, like data versioning, to guarantee reproducibility.
Events
I started to be engaged with the tech community by giving talks and organizing events.
Talked: - At Python Nordeste 2022 (biggest Python event in the northeast of Brazil) about my work on the open source lib Ipyannotator and how it can be used for terrestrial traffic annotation, using the CityCount project as an example. - V PHPRN meetup, returning to the in-person events. I talked about my experience with the TALL stack.
Organized: - The V PHPRN meetup - The IX Meetup GrupyRN - The Python Day 2022 - The PHPeste 2022
It was my first year organizing events and managing the tech community. It is so much more work than I tought initially, talking with the community, designing things, updating websites and social media, scheduling the space, searching for sponsors, ordering the food, and generating certificates. I learned a lot about accessibility, there’s always room for improvement but I think we were in the right direction at the Python Day, with food for everyone (vegan, gluten-free, milk-free, sugar-free), no applauses for people with a sensibility to high sounds, and efforts to have a space that embrace all genders, races, and sexualities.
Exploring
- Learned about computer graphics by developing a simple C++ raytracer that read a
.xmlfile and output a.ppmimage file with a 3D scene. It was one of the most challenging projects for me because I have aphantasia, not been able to draw mental images was a issue to debug camera projections and abstract some angles calculations. - Studied about metrics that measure the quality of projects with object orientation. The Tinylog (Java version) quality was investigated by measuring its Cyclometic complexity, fan-out and other simple metrics.
- Published 4 posts on the Palaimon Blog
- Published a paper about Ipyannotator on the Journal of Open Software (JOSS)
- Took more responsabilities as a data engineer. I developed an intake plugin to help the team to versionate its data and experiments.
- Learn lots of usages and pain of OpenCV by working on the CityCount study. This project was a good test case for the data versioning intake solution and Ipyannotator lib.
- Tested, refactored and extended AI solutions for Multiple Object Tracking (MOT). Most of the AI project don’t test its code which makes difficult to refactor and extend them, for our use case it was necessary to write tests before touching the code, increasing the quality of the software and the understanding of the code.