* 2012/12-2015/06: Making Chrome a future platform for better apps.* 2015/06-2020/03: Making Earth look beautiful (Geo Imagery team).* 2020/04-current: Detecting defective CPUs in Google data centers to eradicate silent data corruptions.

I finally got to work, alongside a long-time friend of mine and colleague in 3 past jobs, at this great company with a noble cause of making the whole human genome sequencing a part of everyday clinical routine in order to improve the general well-being of the mankind. Here, I made a significant impact on one of the core pieces of the puzzle: the software subsystem called genome assembly pipeline, which turned the raw, noisy, disorganized output of the preceding sequencing process into a structured, aligned representation of the patient's genome accompanied by a whole set of statistical analyses and annotations.

I helped to productionize a new generation of Guaranteed Ad Delivery platform -- the core of Yahoo!'s business. I was able to quickly make a significant impact on the product, and received an offer to become a full-time employee, when the position at Complete Genomics finally opened up, so I immediately jumped on-board to that long-sought opportunity.

I got to work on an extremely interesting and challenging problem of automatic test generation for boolean logic circuits, the target being full test coverage of a circuit, as opposed to the currently used tedious manual writing of imperfect and incomplete tests. The startup was quickly acquired by Synopsys, which wasn't the outcome I hoped for, so I left. At the same time, I learned that a position at Complete Genomics -- which I had been eyeing for a long time -- might open up shortly, so I wanted to spend some time contracting without any long-term commitments, to wait and see.

I got burnt-out with the Electronic Design Automation industry and tried my footing at gaming. I worked in the Global Online Studio within EA, improving and expanding the global backend used by all the EA online games. This was one of the best teams I've worked on, but it quickly became apparent to me that my complete lack of passion for games was a serious obstacle, and I decided to try to go back to EDA once again -- this time to a promising startup.

I wanted to switch gears from physical IC design and joined an incubator project called C-to-Silicon: we created a successful tool that translated a very high-level, object-oriented IC description in System C (a C++-based methodology for describing IC designs at a high level of abstraction) down to a low-level, ready-for-synthesis RTL representation, an IC design equivalent of assembly -- the form that IC designers directly code from the ground up today. I worked on the C++ front-end of the tool, applying my in-depth C++ knowledge to significantly expand the subset of System C that the tool supported. One example: I implemented support for C++ templates in the suite (based on Edison C++ frontend).

I joined an innovative, algorithm-heavy, startup-like EDA project code-named Mojave: a massively parallelizable approach to executing design rule checks on physical IC layouts. Performance scaled almost linearly up to hundreds of nodes in the compute farm, cutting down the checking time by factors of 10 and more. I developed several algorithms here and ported the entire software package from Linux to AIX (IBM's UNIX).

I was one of the several computational geometry algorithm developers on a team that worked on a first-of-a-kind flagship tool to aid integrated circuit designers in editing their physical layouts in a much more efficient way. Think of it as an automatic on-the-fly spellchecker and autocorrect for ICs -- guaranteeing an error-free design every step of the way, as opposed to tedious, iterative, manual checking for grammar errors with a dictionary in the hand after the fact. This created a tremendous productivity boost for the designers. We also created an alternative IC editor that was at least 10 times faster than the leading market competitor at handling very large IC designs.

I participated in two large projects at the Lab: HAVi stack and test suite for third-party developers (HAVi stands for Home Audio-Video interoperability) and MHP set-top box (Multimedia Home Platform, a then-emerging industry standard for a smart home network of cloud-connected multimedia devices). Both delivered tangible products, although neither technology got market traction in the end. Working at Philips was a great experience nevertheless: I learned how serious end-to-end software creation process works within a large enterprise and large multi-cultural and distributed projects, and I also picked up embedded C/C++, Java, UML, RUP and many other interesting technologies and tools there.

An incredibly interesting one-year project! We were building a complete hardware/software solution for industrial quality control of manufactured 2D profiles (or thin 2D cross-sections of 3D shapes, such as cables). The hardware part was a light table with an industrial grade CCD camera over it, and the C++ firmware/software suite that our small team of 4 built from the ground up. The suite was able to reliably reconstruct a hierarchical vectorized representation of the profile contours from a noisy rasterized image from the camera, find the best match in a database with reference CAD models, identify the best positional correlation between the model and the workpiece, and perform precise measurements of the allowed tolerances -- all in a matter of milliseconds. I developed several critical proprietary algorithms for this project, mostly in the area of image processing.

Not much of software development here, just support, but I had a great time covering the entire Belarus and half of Western Russia while visiting our numerous branches with business trips, and got a tremendous boost to my troubleshooting and people skills.

Second part-time job: we built complete functional models for several Belarus enterprises using IDEF0 and IDEF1 methodologies (created by DoD) and, based on them, generated complete Informix DB schemas and accompanying UIs to service these enterprises' workflows.

My first part-time job, still in my university years -- nothing interesting, just some FoxPro databases, but it got me started!