After nearly seven enjoyable and stimulating years, July 2nd was my last day at NVIDIA; I learned an enormous amount from all of my wonderful colleagues there, and I'm very grateful for the opportunities that I was given there. Some highlights of my time at NVIDIA include: assisting in or facilitating the bringup of eight (!) chips; designing a ground-up new architecture for Tegra's video input engine; visiting a customer site with a logic analyzer that was worth a few times more than the car I drove there in; laying out the concept of the image processing pipeline for the zFAS platform in this year's Audi A8; prototyping a machine learning-based image signal processor; spending an hour or two with a soldering iron attaching probes to a 3GT/s MIPI link; and guiding the NVDLA open source project from concept to initial release. Perhaps more strikingly, I was lucky to have worked with a great number of people with at least as diverse experience as I, and to have had the chance to benefit from their expertise! So it is with nontrivial sadness that I depart, but that is how things go, I suppose.
On the other hand, it is my joy and excitement to announce that, as of last week, I am now Senior Engineer and Vice President at Accelerated Tech! Accelerated Tech is the combination of me and Jamey Hicks working together to provide hardware and low-level software design and engineering services. I first worked with Jamey over 15 years ago, when I had an HP iPAQ h3765, and his project at HP Labs -- handhelds.org (sadly, now defunct) -- existed to run Linux on those devices. Jamey was a powerful mentor for me during my early days of systems programming, and so I am proud and excited (and, perhaps, a little nervous!) to be working alongside him again in this new role.
When I worked with Jamey last, it was not immediately obvious at the time that what his lab was doing was "the future", but in retrospect, it is very clear to me that the handhelds.org work was the direct predecessor of, and the enabling technology for, modern smartphones. I don't think that it's possible to try to predict future results from past performance, and the kind of impact that handhelds.org had seems like the kind of thing that happens at most once in a lifetime. At the same time, the parallels of stage of development of some of the technology at play are striking: programmable logic has only recently hit a price point that ordinary hobbyists can experiment with, and that it can start to show up in consumer devices. Development for programmable logic is still quite difficult, especially as designs become nontrivial, and it it's still very unusual to build hardware with 'modern' programming techniques (i.e., with higher-order typed languages, or 'correct-by-construction' design methodologies). It's hard to believe that we could have the kind of impact a second time that we did the first time, but there sure is a lot of space in the hardware design world to improve.
I am excited. I think that I'll be doing a lot of different and interesting work soon. I am also a little nervous, for myriad reasons. But that's a good thing, maybe: if the outcome of something is guaranteed, is it really worth doing? No matter what, I hope to learn and grow.