Station A

• Industry: Clean Tech
• Role: Principal Software Engineer
• Date: 2019 - present

Station A was founded when NRG's distributed energy optimization team spun off. I was the first employee to get hired.

My work at Station A covers many aspects: I research, design and improve numeric and machine learning models that enable us to assess a million buildings for their clean energy potential. I also work on infrastructure for data ingestion, warehousing and serving up the results of our evaluations, to our SaaS customers as well as participants in the Station A clean energy marketplace.

Close

Maxta

• Industry: HCI
• Role: Employee #2
• Date: 2012 -

At Maxta I designed and implemented large parts of a data storage solution for virtual machines and containers. The architecture was distributed in nature, with no central point of control and no single point of failure.

The goal was to build a unified storage pool out of off the shelf servers. The system was self-healing and would automatically compensate for storage device or entire server failures.

As the company's second hire I wore many hats...

Here you can see me talk about what we developed at Maxta.

Close

VMware

• Industry: Virtualization Software
• Role: Software Engineer (various titles)
• Date: 2004 - 2011

As VMware started to penetrate enterprise datacenters, customers realized that conventional data protection approaches did not scale in virtual environments. At that time backup vendors were also unwilling to adapt their products because datacenter virtualization was still fairly new and represented too small a market.

Tasked to address the problem, I ended up developing a vendor-agnostic framework for wiring existing backup applications into VMware’s infrastructure.

As adoption of virtualization in the datacenter grew, software vendors were eventually willing to adapt their products and the framework turned into a set of APIs around which an entire ecosystem quickly developed that is still thriving today. Over the years I also made various presentations on this topic at VMworld.

I also spent some time working on VMware’s VSAN distributed storage project.

Close

TBR

• Industry: Software for CAx and Manufacturing
• Role: Freelancer
• Date: 1994 - 1998

I freelanced for TBR during my college years in Austria. TBR developed software for design and manufacturing.

The biggest project I worked on was a facility management application for the Dürnrohr power plant. The software was later adapted for use in other power plants as well and to my knowledge TBR continued to maintain the software until the plants using it were decommissioned.

Other projects were extensions for CAx applications (AutoCAD, for example) that manipulate 3D CAD models.

Close

UNM (aka NetXen)

• Industry: Dataflow Processing
• Role: Software Engineer
• Date: 2002-2003

UNM was a fabless chip company that developed a "dataflow processor": A custom chip design consisting of a large number of simple processing cores with DSP-like capabilities centered around a message-passing architecture.

I developed memory management and scheduling routines for the processor that can be called a simple operating system.

The company was later acquired by QLogic.

Close

Scale8, Integratus

• Industry: Storage
• Role: Software Engineer
• Date: 2001

Integratus and Scale8 were both companies developing distributed storage solutions.

Unlike my work at Maxta, these companies attempted to design a general purpose distributed storage solution that was intended as a replacement for a filer.

Close

Napster

• Industry: File Sharing/Music Service
• Role: Software Engineer
• Date: 2001

Napster was an early peer-to-peer file sharing platform. Its primary purpose was to allow users to share music.

I briefly worked at Napster on the DRM system, when the service was shut down and the company was trying to overcome its legal challenges.

Close

xS+S

• Industry: Custom IT Solutions
• Role: Employee #1
• Date: 1999 - 2001

At xS+S I developed a special-purpose Linux distribution. This also involved bug fixes and enhancements to various open source components, including the Linux kernel.

The distribution that I developed enabled xS+S to provide products such as diskless Linux workstations. The company later specialized in environmental monitoring/data visualisation solutions. For example, xS+S powers the Numbis, the air quality reporting portal for the state of Lower Austria.

Close

Patents with my name on them

US7,606,868, Oct. 20, 2009, “Universal file access architecture for heterogeneous computing environment,”

US7,707,185, Apr. 27, 2010, “Accessing virtual data storage units to offload operations from a computer system hosting a virtual machine to an offload server,”

US7,774,391, Aug. 10, 2010, US8,095,525, Jan. 10, 2012: “Method of universal file access for a heterogeneous computing environment,”

US8,056,076, Nov. 8, 2011, US9,239,731, Jan. 19, 2016, “Method and system for acquiring a quiesceing set of information associated with a virtual machine,”

US8,291,180, Oct. 16, 2012, US8,694,746, Apr. 8, 2014, “Loose synchronization of virtual disks,”

US8,296,759, Oct. 23, 2012, US8,789,049, Jul. 22, 2014, “Offloading operations to a replicate virtual machine,”

US8,443,166, May 14, 2013, US8,954,665, Feb. 10, 2015, “Method for tracking changes in virtual disks,”

US2017-0116302, Apr. 27, 2017, “Replica Checkpointing Without Quiescing,”

US2017-0242751, Aug. 24, 2017, “Detecting and Correcting Silent Data Corruption in a Distributed Storage System,”

Close

Whole Home Electrification

How It Started...

We moved to a new place in 2021. Major appliances were approaching their end of life. We decided to electrify our home.

I wrote this summary, looking at energy usage and carbon emissions convinced us to make the move away from natural gas. Here is a summary of the considerations that lead us to make the move.

...And How It Is Going

Initial data from the 2022 heating season supports our decision from an energy usage as well as from a carbon emissions point of view.

Close

Air Quality

A Python script using dash and plot.ly to find the Purple Air sensors closest to a given location and visualize their sensor data.

This was born out of frustration with the EPA air quality page often being slow to update or not providing any data at all for several hours during the 2018 California wildfires.

Details here.

Close

Bus Schedule

The bus stop outside my house doesn't have a sign. So I built one.

The bus stop outside my house doesn't have a sign showing arrival times. So I built one out of an 8x80 LED matrix and an old first gen. Raspberry Pi I had lying around.

It shows a bit more info than the usual Muni signs.

Details here.

Close

Smart Picture

Dabbling in IoT.

The smart picture is a 7.5" electronic ink display that is controlled by an embedded chip. It is designed to run off a rechargeable battery for about two weeks, while still updating in one minute intervals.

In order to be able to minimize battery drain and to fit into the small footprint of an embedded controller, the data gathering and rendering is done on a Raspberry Pi (that is already serving other purposes).

Also, in order to preserve battery on the smart picture, the code on the Raspberry Pi has to anticipate future requests coming from the smart picture and have a new image rendered by the time the next request arrives, while not serving out stale data.

The project is available on bitbucket. Documentation for the library I wrote to render the glyphs on the display is available here.

Close

Solar Aggregator

Solar Aggregator tracks the energy flow in our home.

When we installed solar panels on our home I wanted to be able to track our net power flow to/from the grid and track our power consumption.

Solar Aggregator gathers production data from the solar inverter and the PG&E smart meter (via a Zigbee gateway) and stores data every 5 minutes in a database. There is a web UI for querying data and for visualization.

Solar Aggregator runs on a Raspberry Pi. It is written in Python and Javascript and uses jQuery and D3 for visualization.

The project is also available on bitbucket.

Close

LAndscaping ManagEment System

LAMES is a hardware/software solution that controls exterior lights and sprinklers. Lights can be turned on/off via a web UI, or according to a schedule. Times can also be specified relative to sunset/sunrise. The sprinklers can be activated according to a schedule as well, with manual override.

LAMES also queries the current weather forecast and it will not run the sprinklers if rain is in the forecast:

LAMES runs on a Raspberry Pi and includes some other hardware components:

Close

Woodworking

Over the years I have also dabbled in woodworking. Some thoughts and a recent project are documented here.