Jan 2013 – Jan 2020 · 7 yrs · Palo Alto

• Invested $184M into 42 companies currently valued at $1.3B (based on exits and last-round valuations by external leads) which is 7x gross MoIC. Realized and emerging hits include Redis, Sysdig, Astronomer, Crusoe Energy, Compound Finance, Tealium, Frame, Gradle, LucidLink, ZeroHash, UMA, dYdX, Cosmos, MakerDAO, and CoinDCX.

Jan 2006 – Jan 2013 · 7 yrs

• Invested $81M over 7 years into 18 companies to yield $481M fully realized, which was 5.9x gross MoIC. Among the 18 were several hits, including MuleSoft, Dynatrace, Lending Club, Buddy Media, SpringSource, Dropcam, Sonatype, Jambool, and Junglee Games.
• Along the way the partnership imploded spectacularly, and in a plot twist, recovered.

Jan 2003 – Jan 2004 · 1 yr

• Spun out my entire original CustomWare team from Borland in 1998 to focus on the next generation of middleware which had Java as a centerpiece, then called J2EE, now called Jakarta EE. We merged with another similar-sized company to become The Middleware Company. After the merger I was President, and later became CEO. The Middleware Company was acquired by Precise Software in 2002, which was acquired by Veritas in 2003, which itself was acquired by Symantec in 2004. I stayed through all the acquisitions, until Symantec's acquisition of Veritas was announced.
• Besides middleware tools and services, The Middleware Company ran TheServerSide.com, a highly popular network of server-side developer community sites, which were acquired by TechTarget.

Jan 1998 – Jan 2003 · 5 yrs

Jan 1996 – Jan 1998 · 2 yrs

• My company CustomWare was acquired by Visigenic Software, a CORBA runtime provider, and I worked for Visigenic for a couple of years, through their acquisition by Borland. My group's role was to make enterprise customers successful with high-performance applications using new middleware technologies, mainly ours.

Jan 1993 – Jan 1996 · 3 yrs

• While at Sun Microsystems and EIT I became enamored with a middleware standard called CORBA, so started a company focusing on tools and services for it. CORBA allowed any software component to call any other software component, regardless of the programming language each component was written in, and regardless of their locations (same address space, different address spaces on the same machine, different machines on the same LAN, or across the internet) without code changes. Later standards such as Jakarta EE and gRPC evolved these ideas for modern times. CustomWare was acquired in an all-stock deal by Visigenic, a CORBA runtime vendor which IPO'd in 1996.

Jan 1992 – Jan 1993 · 1 yr

• EIT was a Stanford professor's startup that self-funded by doing contract R&D for the government while it worked on product, which never materialized. I joined when it was 5 people in my quest to work on the coolest things, and met many very smart people people whom I'm still friends with, but in retrospect it probably wasn't the best career move, and I should have stayed at Sun. HTTP and HTML had just been invented and we participated in the committees that standardized them and fixed the rough edges. Worked on many cool projects for the government using these new technologies. EIT was later acquired by VeriFone.

Jan 1990 – Jan 1992 · 2 yrs

• Designed and wrote a load balancing system which was used internally to entirely simulate two new processors (later called UltraSPARC) at the gate-level, overnight. I was not involved in writing the simulations; my code's job was to distribute them every evening to 200+ server machines of varying specifications that were physically in the building and have the simulations finished and results assembled by the time the hardware design engineers came to work in the morning. I wrote it in C++ and generically enough to distribute any load that took its input from and wrote its output to files.

Jan 1988 – Jan 1990 · 2 yrs

• Single-handedly designed a card with off-the-shelf ICs that plugged into the backplane of HP Vectra x86 workstation prototypes, and could be used in place of more expensive logic analyzers to debug the logic and timing of the workstation's bus and other accessories. Logic analyzers were scarce, even though HP itself manufactured them; hardware design teams had to share them. That made this a perfect independent nice-to-have project for an intern. The card made it possible to do a lot (but not all) of the observation and debugging without the logic analyzer. I had the card contract-manufactured in small quantity, wrote the device driver for the card, and also some basic application software that visualized data.

Jan 1986 – Jan 1988 · 2 yrs

• Wrote software for sensors to measure leakage inductances from electronic ballasts (the circuits that drive fluorescent lights) to make the ballasts more efficient. Debugged analog circuits for flash analog-to-digial-converters (ADCs).