Eugene Kirpichov

This is a proof-of-concept model-checker for concurrent systems (it proves the concept that a model-checker is actually very easy to develop - and this proof has inspired at least one person I know to develop a modelchecker of their own for production-use purposes).
For now, the modeler core is complete, and verification of CTL properties. Many other things wait for their day to come - LTL verification and a friendly front-end.

Written originally for the cluster computing project at Mirantis.

My student's masters thesis. Views, merges, filters etc. unlimited-size logs.

http://github.com/jkff/timeplot , http://github.com/jkff/splot

I wrote these two tools for analyzing behavior of large distributed systems that I deal with at work. One (splot) is for drawing a "birds-eye view" of the whole system's activity, the other (tplot) is for plotting all kinds of quantitative characteristics.
They help me answer nearly every question I have about the system's behavior by piping the logs through a shell one-liner, and I use them dozens of times per day. My… Show more

http://github.com/jkff/timeplot , http://github.com/jkff/splot

I wrote these two tools for analyzing behavior of large distributed systems that I deal with at work. One (splot) is for drawing a "birds-eye view" of the whole system's activity, the other (tplot) is for plotting all kinds of quantitative characteristics.
They help me answer nearly every question I have about the system's behavior by piping the logs through a shell one-liner, and I use them dozens of times per day. My co-workers and ex-co-workers use them too. Show less

My student's masters thesis. Similar in spirit to Google sawzall, but in a streaming monitoring flavor rather than batch.

Powerful yet lightweight highly scalable solution for distributed parallel computations. Created for running computational algorithms on large clusters (tested on 5000+ cores, estimations prove the architecture to be efficient on clusters of up to 50’000 cores). Project took 1.5 years (including 8 month of Proof-of-Concept phase), was successfully released and delivered to customer.

Project Role: Tech lead.

Written to (successfully) optimize the build at Yandex.

An incremental abstract-algebraic regular expression engine for Java .

See article: http://jkff.info/articles/ire/

Several contributions to the stream-lib project by Clearspring (now AddThis) - a library implementing various data-stream algorithms (algorithms for approximate processing of vast amounts of data). My contributions so far: Q-Digest, Count-Min Sketch, general cleanup/optimization.