Eduard Kieser Profile

Eduard Kieser - Professional Profile

Engineer
Berlin, Cape Town

What would I want from the perfect job?

This is a question that I often see on job sites, so let me take a crack at it here.

  1. Engaging work that lets me enter a state of flow. I'm told that
    • freedom
    • variety
    • clear tasks and
    • feedback
      are key ingredients to engaging work
  2. Work that helps others.
  3. Work that I'm good at, or work that I can get good at.
  4. Supportive colleagues.
  5. A job that fits my personal life. I don't like commuting regularly, so working remotely is very important to me. I also appreciate having some amount of flexibility in the hours I work.

The above is a little generic and also shamelessly plagiarized from 80000hours.org.
Lets see if we can make it a little more specific by reviewing my career up to this point, and commenting on what I liked and disliked.

Work history

2015-2018: The Early Days

My work life started in earnest in my final year of grad school with an internship at the Philips Patient Care and Measurements research group. It was a nice experience, but things got really interesting afterward.
The end of 2015 until August 2018 was an awesome blur. In this time I was an independent contractor working as a software engineer, (failed) start-up founder, design thinking coach and medical researcher in asynchronous bursts.

Eduard the start-up founder:

This is the role that I assumed most frequently in this time. It started in late 2015 when, as a contractor I was asked to do a piece of work for a client, which involved ICU data. I quickly realized that the world was broken in a messed up way, and that this broken-ness was causing serious harm. I did a longer write up on the project here, but the abridged version is as follows:
TL;DR:

  1. Data collected by medical devices is locked into ecosystem silos by companies hoping to profit maximally from it from the moment it is recorded.
  2. Most systems in use in the developing world are considered unsupported legacy systems and there is no practical way to integrate this data into patient care (even if the institutions could afford the exorbitant license fees).
  3. Almost all medical devices have a screen to report the present values of what they measure. I built a system that exploits this feature using optical character recognition, to allow the users to stream and log the data so that it may add value where ever it may be relevant.

This was an amazing time in my life, days of deep focus would fly by and the mission combined with the potential impact of the solution meant that I was extremely driven and quite productive in this time. I was also fortunate enough that my consulting work, (which was also super cool) was able to sustain me while I did this. There were of course also some things I would do differently the next time:

  1. I was too alone. Despite the occasional helping hand from my girlfriend (now wife) and some excellent friends, I was alone for most of the journey with a rotating cast of characters that came and went as opportunity presented itself and interests aligned.
  2. Not enough networking. For most of the project I spent all of my capacity improving the code base and prototype. I figured that if I can show a prospective investor a really good prototype that the project would speak for itself. Because of this approach I still have not shown this work to a legitimate potential investor.
  3. I was naively swimming among sharks. I recognized early on that I was not a good salesman and that I needed funding. I mentally delegated this to a trusted local organization and got screwed over pretty hard. Eventually I had neglected the consulting side of my operation too much, was burned out and forced to get a normal job.

Eduard the Design Thinker:

I was fortunate enough to visit the d-shool in Potsdam in 2013 for the first time as part of the GET program. The next year the GET program was looking for a part time coordinator and facilitator and I fit the bill. Through the GET program I was fortunate enough to be invited back to Potsdam, eventually as a coach and was able to hone my skills as an innovation facilitator at the Hasso Plattner Institute of Design Thinking in Potsdam. I was later one of the founding members of the d.school in Cape Town and was part of the teaching staff for the very first basic track that ran there. This was a really cool part of my life. The work was much less technical and way more social than the other things that I was involved with and has had the greatest impact on how I interact with and relate to people. It was one of the times that I stepped out of my comfort zone, only to discover that my comfort zone was bigger than I had thought and I am really grateful to Marco, Claudia and all the rest that had a hand in making this possible for me.

Eduard the Researcher:

I did a few research projects in my time as a consultant, the biggest that I was involved with was the Safe Passage Study, with the excellent Prof. Hein Odendaal. I was tasked with building a framework that could:
1. Extract data from a closed binary format to a usable open format (at scale, ±10k participants with 1hr of ecg/emg data each).
2. Generate features from the raw signals that could later be used for hypothesis testing.
3. Evaluate certain hypotheses.

Part of my role was also to help formulate arguments, present findings (figures, graphs ect), and help with writing our research papers.
This is another one of the situations that I'm really grateful for. The work was interesting and the feedback was always filled with gratitude and positive reinforcement. I don't have a write up dedicated to this work, but most of the publications listed here are a result of this project.

2018-2020: The HealthQ Days:

When I joined HealthQ I'd settled into my new identity as a full time data scientist. For the first couple of months there was not a lot of freedom and there was a lot of work to get done. Fortunately I had a lot of excellent colleagues to learn from, both technically and also how to handle the pressures of life in a start-up. Even though the company at the time was already 7 years old and had more than 50 employees, it still operated as a start-up with all that that entailed and it was quite the adjustment to the philips environment which I had been exposed to years before. After the first couple of months there was a lot more freedom and opportunity to do cool things. By the end of 2019 I was co-responsible for our project that aimed to make pre-consultation cardiac monitoring and reporting mainstream. More here. I really enjoyed the freedom of being responsible for an independent project, as well as the clear objectives that go with it. By February of 2020 we were crunching the data of our first trails, with really promising results and by March the company had shut down.

While I really enjoyed all the colleagues that I interacted with directly at HealthQ the signs of malcontent and non-conducive pressure from above were there from the start. Seemingly illogical decisions made by people that I've never met regularly dictated what I had to do next and my only way of dealing with it was to just disengage and do as I'm told. This was a huge shame in my opinion, because a passionate, invested, sometimes disagreeable engineer is far more likely to contribute to true innovation than a subdued automaton.

2020-2023: The 42air Days

This was another really lucky break for me. Shortly after HealthQ closed down, one of my friends that worked there with me told me about this california based drone start up that was hiring engineers in Cape Town. I applied and was the second employee of the CT branch of 42air. I started out as a machine learning engineer, focussing on embedded vision systems, specifically depth estimation, pose estimation and object detection. This reminded me of the pre-HealthQ days, the work was fully remote (since there was no office) and we had a lot of freedom. The team grew organically from our own contact list and I had the privilege to spend 3 years working with some of the best (if occasionally most pedantic) engineers you could hope to encounter. In this time I also learned some of the lessons of a true start-up. The most salient was that your title is irrelevant and that you do the work that needs doing. Some additional challenges that we had to overcome were related to the fact that we pivoted to have a greater focus on hardware within the first few months, while still not having an official office. I'd moved to the suburbs by this time, so I was fortunate to have a little space and my place became the defacto office/workspace for the first few months. This was an awesome time and I have really fond memories of this. Part of being a small team of exclusively engineers, meant that there was a lot of menial admin that was being done by highly skilled and well payed engineers that were not very good at this sort of thing. By 2023 we had gotten some support staff in Cape Town and started small batch (10 units) production runs of our "gadget" hardware package. Drones using our hardware were doing regular delivery flights to customers in the mississippi river. The last time I checked our servers we had done around 600 flights about half of which were for paying customers.
In the start of 2022 we started the BlueCrane program. The BC was our home brewed VTOL to expand our payload and range capabilities for more challenging use cases. By this time our team was spread pretty thin and I did most of the work on this project alone. This started with settling on the specifications with our CEO and included:

  • Writing software to quickly visualize and iterate over airframe concepts and layouts.
  • (With the help of Andre our excellent intern) Developing a hardware test rig to accurately characterize our power system equipment. With heavy lift VTOLs the margins are really fine and the spec sheets that are provided by even the most trusted brands are not sufficient to accurately characterize expected performance.
  • Writing software to use the results from the hardware characterization system to simulate the performance of various configurations for different mission profiles.
  • Designing non-standard electronic subsystems (we had many many sensors on board).
  • Designing the airframe.
  • Manufacturing capacity planning.
  • Leasing with contractors to build the components that I could not make in my workshop (carbon fiber fuselage and wing sections).
  • Assembly and integration.
  • Flight testing 😎.
  • Making the system that generates the report for the flights based on the captured flight data.
  • Taking this data and making decisions on what to do next.

During this time I also contributed to a few open-source projects including the ArduPilot project, not as a code contributor, but by filing bugs, suggesting features and testing and reviewing experimental branches, stuff like this.

As you can probably tell, I really enjoyed this work and once again allowed it to consume most of my mental space. BUT. I would have been far better of if I had managed to divert one or two team mates to the project. I only reported on the very high level progress that was made, so there was no one to appreciate the challenges that were overcome in order to deliver the terse updates.
Doing work that helps others is not a nice to have, but a requirement for a happy work life. Working alone rarely gives you this opportunity, I would do it differently the next time round.
In 2023 our parent company changed ownership and our little operation was deemed not to be compatible with the new owners greater vision, so 42air closed down in August 2023.

Present Day

I'm once again in the position where I need to figure out what needs to happen next. I think I'd be happy in any role that aligns with the criteria I've mentioned above. If that's to vague then perhaps consider the following examples.

  1. ML for health care. I believe that there are a lot more low hanging fruit in the health care technology space. Especially when it comes to adapting existing solutions to work in resource constrained settings, where these solutions are often needed most.

  2. Robotics for agriculture and logistics. I love things that fly and I love it when robots really nail a particular role. There is still a lot to be done in precision agriculture and I would love to be one of those that take on the remaining challenges.

  3. Renewable energy. Having a robust carbon neutral energy supply is a requirement for human kind to succeed. Seems like a worthwhile thing to spend one's time on.

  4. R&D in general. Making things possible/practical that were previously thought impossible/impractical.

Skills and Technologies:

Throughout my career I have learned to use many technologies and tools to varying degrees of proficiency.
In my ideal situation I would probably rely heavily on some combination of these technologies and resources:

  • https://arxiv.org/ and sci-hub
  • Python, Docker, Bash, Git, Go, C/C++
  • PyTorch, TF, ONNX
  • Pandas, sk-learn, openCV and friends
  • Ardupilot, ROS, ZMQ, Mavlink
  • Flutter
  • AWS ECR, EC2, S3, Sagemaker ect.
  • Fusion360, Autodesk Inventor, KiCad
  • Stack Exchange, Stack Overflow, ChatGPT and Google.
  • New things, perhaps, Mojo and the Modular framework (very keen to try it out).

Working remotely

The ability to work effectively, while working remotely, relies on a skill-set that needs to be learned. This includes:

  • The ability to communicate effectively and asynchronously through documentation.
  • The ability to make decisions independently and discerning when to involve other team members.
  • Mindset of self-enablement.
  • A whole lot more.

For the last 3 years I've been working from my home in Cape Town, for a company that was headquartered in San-Francisco. The vast majority of my team were also working from their homes, some in the US, some in Cape Town and one in Perth.

The primary challenges of this arrangement stem from the administrative complexities of legal and tax compliance. No company wants to establish and manage a local legal entity just to hire one engineer.

To streamline this, I've completed the necessary legal groundwork to enable prospective employers to engage with me via a single contract, covering remuneration, responsibilities, and IP rights. For you, this would be a B2B interaction, considerably reducing your risk.

So why not be a consultant? Although what I've described might sound like a consulting agreement, I'm seeking more than just a contractual connection. I want a sense of belonging. I aim to be an integral part of a team, share in its long-term vision, and anticipate the same public holidays. Hence, my ideal engagements would use 100% of my capacity and be of extended duration, with a default termination date set for years in the future.