I started working at Lawrence Livermore National Lab in the late spring of 1998. It's was my dream job: ever since I was a kid I've dreamed of being one of great minds that work at the National Labs. Dr. Feynman has been one of my greatest influences. Unfortunately, particle physics is not that great of a career choice right now (I was at Fermilab in 1993 when congress killed the SSC: you could cut through the dissapointment with a butterknife). I ended up in Computer Science, which is not only a good career move these days, but a field that Dr. Feynman had a great interest in as well. The period in history that I would most like to see is Los Alamos during the Manhatten Project with their punch card driven calculating machines.
So I ended up at LLNL, which is really the place to be for Computer Science in the National Lab system. Okay, so maybe I'm really pulling for my team in the little intralab rivalery that goes on, but it's all for the good of science. Anyway, I wouldn't want to live in a desert and Sandia-Livermore is just so small. . . Of course, everyone knows that the three defense labs are the DOE's babies (if I don't get flamed for that statement, I'm going to know that people aren't reading this page).
For my first two years at the lab I supported what was called at the time I left The Center for Computational Engineering. My primary job assignment was to develop TIGER, which was a set of technologies (which materialized in the form of a library) to abstract the bookkeeping from the physics in scientific computations, particularly time domain electromagnetics. It was a success from a computer science standpoint: the core abstraction (the Entity/Attribute model) is particularly compelling and I did some really cool work with abstracting away the node that a particular Entity lives on in a massively parallel environment. Not to mention some fun stuff I did with multiplexing MPI communication with threads. The problem is, we weren't able to polish it off before funding ran out and it never got physics in it to prove that it's actually useful for a real application. My last assignment in Engineering was working on a project to develop some solvers for use in various EM analysis codes. The problem that I looked at specifically is that these solvers are all in a nice C++ framework, but we need to be able to call them from Fortran 90 applications. So I created Interfacer, which is like SWIG in that it creates interfaces to C++ classes so that other languages (Fortran, in this case) can call them.
I left engineering to work in the Cyber Solutions group of CIAC. The Cyber Solutions group creates tools (or solutions, if you will) that assist users and the rest of CIAC keep their machines secure. My first job at CIAC was porting a tool called SafePatch to Linux. SafePatch allows administrators to centrally administor OS patch installation, hence ensuring that all machines in their department have the most recent patch levels on them. After SafePatch I started working on Logger. Logger was a program that collected log data from various sites and applied data mining techniques to it in order to find interesting computer security information. When I started on the project, I figured we could just take some off the shelf tools to do that sort of thing. After a careful search, not only did I not find anything, I found that it was a very active area of research. As such, it became the focus of my Ph.D. research.
While my research continued, I was given the opportunity to work on a very interesting problem in information fusion, which utilized massive-scale graphs to do knowledge management. When I say massive-scale, I mean millions or more pieces of information, all linked together. I worked on that for five years, under different sponsors, ocassionally doing some computer security work when the funding dried up, eventually becoming the senior scientist, and ultimately project leader for code development until Congress pulled the plug for political reasons. After that (and actually, for a bit before I became senior scientist), I worked on the Predictive Knowledge Systems (PKS) research project, which was essentially the research into what we can predict, given such a massive-scale graph, fusing in additional (temporal and geo-spatial) data.
It was durring this time that a small government contractor called Eyak Tek found my resume and (correctly) thought that I'd be perfect for a new contract that they have with NASA Ames Research Center to do network flow monitoring -- essentially the same thing I was doing on Logger back at CIAC. So in early 2008, after nearly ten years at LLNL, I left to become a "Senior Intrusion Analyst Engineer" for Eyak, working down at Ames. I'm going to continue to link to this page as "Global Thermonuclear War", both because it's fun, and because without the research from NASA, there wouldn't be a delivery mechanism for the bombs (Ames, in particular, houses the premiere wind tunnel testing systems in the world).
|Biography||Contact Information||PGP||Projects||Geek Code||Audio Visual|
|The Meaning of Zow||Global Thermonucular War|