I started my professional career almost 15 years ago working for a Machine Learning / Computer Vision company that did image recognition on mobile. At the end of 2013, I chose to leave it for Lima, in part because even at Moodstocks I was always more of a Distributed Systems / Algorithms person than a Deep Learning person.
Now, after 10 years away from the field, I am back working on AI for images, this time more specifically on automated and realistic image edition, with a part of the old Moodstocks team who spent years at Google after the acquisition.
I did not make that choice lightly, I spent some time pondering it, because I knew it would require a significant time investment. AI is a complex, fast-moving field where not everything you learned elsewhere applies. Now that I am back, it is unlikely that I will leave it anytime soon.
I got several interesting job offers recently, for Staff+ roles at great companies with classical distributed systems problems that would have appealed to me a few years ago. But it is likely that the real dist.sys fun in the next few years will be in AI, both for training and inference. The field has grown fast, and the tooling and practices could not keep up so there is a lot to build.
Those who know me may also remember I always had a problem with the issue of interpretability / explainability of deep learning models, but in recent years I feel we have collectively improved on that, although there is still much to be done.
In addition to that, I do not intend to leave Finegrain specifically either! We have a great team and we approach problems differently from most AI companies. We have products starting to ship, and a lot to do to improve them. I don’t think I should be anywhere else.
Anyway, I am taking the end of the week off to attend and help with the dotAI conference, and I will be back to work next Monday. In the meantime, try the Finegrain Editor if you haven’t already :)