Why robotics is changing everything right now | Clemens Marschner

00:00:00: Welcome back to the Robco podcast.

00:00:02: And today we have a special episode where

00:00:29: Physical AI.

00:00:30: No theory, just

00:00:32: reality.".

00:00:35: So welcome Clemens.

00:00:36: and yeah please introduce yourself um... And tell us a bit about what it means to be a principal engineer at Robco?

00:00:45: Well first of all I'm probably the oldest engineer so that's one thing that it means!

00:00:51: Um..I have history with um.. AI because before it was called AI or before you were able to call an AI I was working for US West Coast companies like Microsoft.

00:01:04: Um, I was in the ride-sharing space and I worked on autonomous driving a couple of years always out of Munich.

00:01:11: Oh right!

00:01:12: And so RobCo came about to give me that opportunity to work at the headquarters of a company which kind of works as Silicon Valley startup.

00:01:23: Cool what got you interested into this space?

00:01:25: Like looking at your school year.

00:01:26: What's interesting is technology.

00:01:30: I've been a computer kid, uh ever since computers came into my vicinity.

00:01:36: So My brother got a Commodore sixty four when it was ten years old and since then i was hooked.

00:01:43: It got to the point where after finishing school?

00:01:47: II thought you know this is all too boring.

00:01:50: I don't want to study computer science now.

00:01:53: And so then I did it, did a little detour and then came back.

00:01:57: Okay that's probably the only thing i'm able to do.

00:02:02: Did you ever use Atari or what was your path through different computers?

00:02:08: First Commodore as I said and then Amiga.

00:02:11: The classical eighties and nineties career in that sense at some point on PCs.

00:02:18: How Yeah, how did you go through the two thousands when there was the big dot com bubble?

00:02:24: or at least they're big internet shift?

00:02:26: Did You like what with your first touch point of the Internet.

00:02:31: My First Touchpoint Was At TUM at The Munich Technical University where They had an Open Door Day and so I was able to get There.

00:02:43: it was Fallen Of Ninety Five And I typed in an address like yahoo.com and it downloaded stuff from the United States, and that was just completely thrilled by that!

00:02:58: That is such a crazy technology shift right?

00:03:01: It's not that long ago.

00:03:01: we're talking about thirty years now of working internet let's put it this way.

00:03:07: so you studied at TUM at TU M Technology University of Munich.

00:03:14: what field did they study then?

00:03:17: Actually, I was at the LMU.

00:03:18: Oh okay!

00:03:19: So i shifted a bit after... Like that was before finished school and eventually studied Computational Linguistics

00:03:32: Okay

00:03:32: And did a PhD there while already working as software engineer.

00:03:37: Interesting!

00:03:38: The core always self-taught in everything im doing.

00:03:43: so it kind of took awhile Like non-linear career path, but with the PhD eventually I settled on what would be doing.

00:03:54: How did you figure out?

00:03:56: What to do after school?

00:03:56: because as we're talking about a non linear path that is good thing too.

00:04:01: figuring it's the hardest part right.

00:04:03: Afterschool especially in Germany.

00:04:05: just picking something and going into studies Is easy or not easy But its fast And then can't go wrong.

00:04:13: So tell me your thought process Tell me a bit about the non-linear path, because I think it's really interesting to be figuring out what you're passionate.

00:04:23: Right?

00:04:23: So here is the complicated part.

00:04:24: so...I was always talented in that space but..it wasn't cool back then!

00:04:32: It was before people like Mark Zuckerberg come out of the gym every other day and so on fringy, geeky nerdy space.

00:04:44: And I didn't really want to be part of that and so i was actually looking for ways.

00:04:52: you know where are fields?

00:04:54: Where there more girls and uh...where you can have more fun.I had this other interest in the media industry.

00:05:01: So I looked around at that space a little bit and eventually um..you have make choice whether You want to be mediocre in a field that interests you,

00:05:12: or

00:05:13: if you wanna really good and up into place where have the talent.

00:05:17: And so eventually it kind of went back too bad when your successful then also start.

00:05:25: No,

00:05:26: that makes absolute sense.

00:05:27: And it's interesting.

00:05:28: I said the word passion before and actually i think That It Does Make Sense.

00:05:32: You Like The Stuff You're Working On But I Think Passion Has Been Misused A Lot In The Past.

00:05:38: When You Recognize You Have Talent At Something you Can Work in That Space Then It Becomes A Lot Of Fun As Soon as You Have Success.

00:05:45: So I Think Its Always a Mixture of Both.

00:05:47: Having This Because Sometimes You Just Have Talents For Technical Stuff Otherwise You Have Talent for Creative Stuff And Sometimes Those Are Even Mixed.

00:05:55: It's really interesting to see that.

00:05:57: And then you studied at LMU first linguistics, computer linguistics?

00:06:04: How was that?

00:06:05: I mean... That must have been a crazy topic right?

00:06:07: Yeah so i came through the topic of search engines.

00:06:11: So back then search engines were huge thing.

00:06:14: it was time when Google came about and search engines before where like pretty useless You could find a million documents but ones that are relevant didn't come up.

00:06:25: And that's at its core, it's a language understanding problem and obviously also has to do with you know clicks on all of these days but the internet was young so there wasn't really lot data around.

00:06:39: then I got dragged into this whole area of language processing not being a linguist myself.

00:06:49: pretty quickly came um, machine learning or for language processing which was a growing field at the time.

00:06:57: Um so this got me into the whole area of you know statistics and and machine translation and all these kind of things.

00:07:07: I mean basically machine learning is The logical path before AI artificial intelligence And so that put you in to very early stage Of understanding how How that whole field works right?

00:07:20: Early and late.

00:07:21: I mean, language processing is one of the oldest fields off artificial intelligence.

00:07:26: right since the nineteen fifties people have tried to build machine translation systems for example as like that The primary task to solve um And they've been trying for decades.

00:07:39: Um Eventually, when I went to conferences like in the early two thousands they were these fifteen different tasks and everybody was working on a tiny part of this problem.

00:07:54: And all that got solved over the past few years but the whole field has just completely revolutionized for the last few years.

00:08:03: now i think we're at stage again where you can with good conscience say okay we are an age of AI

00:08:10: Amazing.

00:08:12: Then looking at your PhD, then you chose a PhD field?

00:08:16: How did it choose that and how did that happen?

00:08:21: now I need to remember what actually there is long times like several career switches ago.

00:08:29: again i came from the area of search And we were looking into very deeply into some semantics of texts.

00:08:39: So my PhD advisor was not a machine learning guy, he was coming from the logic field that had written like seminal books about logic and so for five years We were arguing because He asked me to do The PhD there Like I didn't apply.

00:08:52: if i If i Had known how this whole thing works?

00:08:55: I would have probably gone To maybe Maybe A different place.

00:08:58: but so This way We were arguing, I was arguing has to be solved by statistics but data-driven approaches and he was arguing against it.

00:09:06: I think eventually i was right.

00:09:07: that's fair to say um But it was kind of healthy... ...I learned a lot about the whole logic field And then eventually you know finished the PhD and never touched logic again.

00:09:20: And after creating your PhD, when did you go to Microsoft?

00:09:23: Was that before in between or after.

00:09:26: That was around the time When I finished it.

00:09:29: The company I was working for was acquired by Microsoft

00:09:31: makes sense.

00:09:32: and so i said Oh shit!

00:09:36: The topic I'm working on Is is In connection with a Company I was Working With.

00:09:42: So now management changes.

00:09:43: so better get this out Um, so that you know because priorities might shift and then I'm gonna have trouble.

00:09:50: And so on.

00:09:50: So i finished it around.

00:09:53: uh two thousand eight um and then switch teams.

00:09:57: Uh was them working on web search ranking for quite a while.

00:10:00: interesting very interesting.

00:10:02: It's also interesting we'll find out soon enough but how the switch came from your Web Search to robotics Different different.

00:10:11: What is your next step thing?

00:10:12: reason?

00:10:13: Uh, what was the next step after Microsoft then?

00:10:16: Um,

00:10:17: Microsoft at some point um divested a bit in web search ranking because it was felt that The big leaps had already been done and It was like it was Bing at that time.

00:10:30: so it was kind of comparable to Google And so became a profit center.

00:10:36: Our team looked around inside of the company for new topics.

00:10:41: And that took like one or two years, I was working on some interesting projects around deep learning computer vision and also the mixture between computer vision an natural language processing which both images and text to the same currency, which is just vectors.

00:11:08: So now suddenly you could build things like image captioning models And so I was very much hooked into that.

00:11:16: but there wasn't really a field emerging where i could do that work especially out of Munich.

00:11:21: um...and so eventually I was contacted by a recruiter.

00:11:25: they said hey we're building an autonomous driving team in munich.

00:11:29: We have core technology called simultaneous localization and mapping, which we're building a team around.

00:11:38: Are you interested in being working on ML in that field?

00:11:43: So I got contacted.

00:11:44: there's bunch of people from Google came here so it was really interesting.

00:11:48: We were building the team at Munich kind-of fit And then I worked on autonomous driving.

00:11:54: How is time working on autonomous driving?

00:11:57: because I think that field, obviously now we're also really advanced in that space.

00:12:01: It feels at least like it when close to self-driving cars.

00:12:05: but that feel must have been super futuristic as well?

00:12:09: It was very much peak hype back then.

00:12:12: so there was this feeling you could commercialize self-driving cars soon and turned out actually, in the end much more expensive than people thought.

00:12:25: So

00:12:26: that's I think... That's a huge thing.

00:12:28: maybe you can clear something up?

00:12:31: Exactly agree with what he just said!

00:12:34: It must be not easy but it must be NOT THAT EXPENSIVE.

00:12:37: Why is so

00:12:37: expensive?!

00:12:39: It's always the long tail right?

00:12:42: You can get ninety-nine percent accuracy with everything that your doing.

00:12:47: and then there are the squirrels You know, going over the road and then some.

00:12:53: suddenly everybody breaks.

00:12:54: And you notice.

00:12:56: I think self driving is pretty close to AI completeness.

00:13:01: so you have to solve all of AI in order to make it work.

00:13:05: now we can.

00:13:05: We Can Make It Work just by throwing in tons and tons and tonnes Of data and really weird situations.

00:13:11: So that's how we are working around That

00:13:13: okay?

00:13:14: Um

00:13:15: and back.

00:13:16: Then it was also Still Classical Robotics.

00:13:20: So, so I think if you look at the transition from rule-based algorithm based approaches to completely data driven approaches.

00:13:33: We had natural language processing making that switch in the past few years.

00:13:38: we have computer vision making it switch and mid two thousand tens robotics was a bit late.

00:13:45: actually what people did is they would build algorithmic planners, which do some kind of optimization and they try to optimize the path that you drive.

00:13:55: But then a lot of stuff around it special cases were written in code

00:14:00: okay?

00:14:00: And so even also for us when we started didn't have a lot data So a lot Of coat was written and eventually you have a planner.

00:14:06: That has like one-and-a-half million lines off C++ code.

00:14:10: It becomes just unmaintainable and you only covered Like A tiny sliver of what reality is all about.

00:14:18: And so eventually you need to make that switch, throw out all the code and learn everything.

00:14:25: Eventually things get better.

00:14:26: but it's kind of a hump because first think they're there then the bar is actually much higher than your thing.

00:14:33: I finally understand.

00:14:33: So basically not The car would work in ninety nine point.

00:14:37: nine percent Of their cases.

00:14:38: That what everybody was able To achieve at let me say pretty fast.

00:14:43: But then last one percent or zero point would kill the business, obviously.

00:14:48: One big accident in the press and media would definitely hurt the whole system And people will lose trust.

00:14:54: That's where it gets so difficult So long-tail as you just said to finish that problem.

00:15:00: I also understand now.

00:15:01: of course AI can solve for unlimited issues at some point.

00:15:07: It'll always have an answer if something happens In a very fast manner.

00:15:11: Eventually build what is called data driven organization.

00:15:14: So you get into the business of making sure that you have sensors out there, that can collect data.

00:15:21: That you have a machine for filtering out non-interesting data curation pieces where human has to look at certain data points and then eventually turning them in models with good metrics when you apply it on new settings so they can give your prediction about how the model performs.

00:15:43: And that's something we did already in Bing very early on, it was the first search engine as an answer to Google which is completely machine learning based.

00:15:53: It had a good methodology for doing this and so ever since you get into new problems with different fields but can apply the same techniques or methodologies.

00:16:04: Very interesting.

00:16:06: So what happened after autonomous driving?

00:16:11: So first of all, the autonomous driving was a pretty weird ride in the literal sense.

00:16:18: It was team that was bootstrapped and then went through an exponential growth curve from zero to three hundred people within like twelve or eighteen months And there were peaks.

00:16:32: but it kind of lingered when looking for partners didn't find good industry partner it got sold off to Toyota.

00:16:42: Interesting,

00:16:43: and so that was like four hundred fifteen million dollars invested.

00:16:49: yeah um... And then eventually I again wasn't in the position.

00:16:54: do you now move to the US or UK?

00:16:57: Or did find something in the company to work on?

00:17:00: The interesting problem where there is a big need was building kind of the Google Maps.

00:17:10: The company was Lyft as a ride-sharing company, is main competitor of Uber and so they were paying a lot money to Google for using their maps.

00:17:21: but they had like a million cars on the road all the time sending their position in there velocity into servers every second.

00:17:29: So that's an interesting problem for somebody who does machine learning because now you have almost infinite data pipe and now you want to make sense of it.

00:17:40: And we were building models for doing the routing, for doing traffic prediction... For showing green, red or yellow on the map obviously very close to business because every routing decision, every ETA estimated time of arrival that you compute has an impact on the business.

00:18:04: Are you making a profit or not?

00:18:05: So it was kind of job security and after three years going through this roller coaster, I was actually pretty happy to find a place where there's no lot companies out here that do these so they don't get competition from all sides.

00:18:21: It is very interesting problem.

00:18:23: i finally got know how these map applications work And at some point Now I know how it works.

00:18:32: No, I helped the business quite a bit and what else is out there?

00:18:38: And then i looked around eventually...

00:18:40: I love that you always look at problems You want to solve and your not looking for or say you're Not Looking For A Job!

00:18:51: You never mentioned any job Or role..You Always Mentioned The Problem That You Thought Was Very Interesting That You Wanted To Solve.

00:19:00: It's very interesting that you see the world like that.

00:19:04: Yeah, it is really inside of view... ...that we want to solve problems with futuristic problems as well and be ahead-of-the-curve in these fields.

00:19:14: I find this fascinating because it means an intrinsic motivation for us to advance technology.

00:19:25: The nice thing about our field of technology that it's so relevant and the demand is so high.

00:19:35: And you can actually make money just by going out, trying to be useful.

00:19:43: So if I'm a screenwriter or an opera singer... Of course!

00:19:50: That's as good of a profession but much harder than making a living.

00:19:59: You just mentioned something really interesting, and I think you had the decision several times in your career.

00:20:04: Staying Munich or move to a different country with different city maybe even Silicon Valley itself?

00:20:10: And my question is there why did you stay in Munich?

00:20:14: Why what is?

00:20:15: why is Munich such place for you where he wanted to stay?

00:20:20: In the end it's humans around you right also of course i think munich Is one Of The Best Cities In The World To Live.

00:20:27: I speak the language, even speak to dialect.

00:20:30: So that lends itself... ...I think it's kind of a privilege to live where you've grown up.

00:20:35: but there was also things like my mother got very ill when i was just in twenty-twenty three and so uh.. That was a component as well.

00:20:48: then had girlfriend and wife um.... These were more important for me than prioritizing career And so eventually that was just for me kind of an easy decision, you know optimize your job given these constraints.

00:21:03: Yeah and I think Munich is just... You know if it were some other city It would have been much harder.

00:21:10: but Munich has the right environment and i think that environment has gotten better over the past years.

00:21:17: its a more international city now than ten years ago.

00:21:21: Oh absolutely!

00:21:22: I think that, and i've moved here twenty years ago.

00:21:27: And I think Munich has changed from being a German city to be one of the most international cities in Germany.

00:21:36: Obviously this helps because fantastic companies are there right?

00:21:39: So you mentioned Microsoft has a headquarters here.

00:21:42: Google has a headquarter here.

00:21:44: There's dozens other international companies open.

00:21:47: A.I just opened their headquarters And I think that makes it a very special, unique place where you can actually not only live a very amazing life because the nature is so nice but also have unlimited career possibilities in an environment such as Germany.

00:22:05: Yeah and just the environment... You have mountains with very nice surroundings ...you are in Italy for two hours.

00:22:15: original Italian espresso there.

00:22:18: You have an airport, which is one of the largest in Europe.

00:22:21: you can get everywhere on vacation here in the middle of Europe so you can't get anywhere within Europe with like one or two at most three hours.

00:22:29: So that sense we're just very privileged here.

00:22:33: I agree.

00:22:34: What was your step after finding out how maps work and solving problem machine learning?

00:22:41: I

00:22:42: was looking around, thinking about forming my own company.

00:22:47: There was a time when ChatGPT came out... ...I actually had a hobby around construction and architecture which is something that got really interested in.

00:23:03: it's also field where productivity has stalled gotten worse over the past decades.

00:23:13: So I was really looking around for, you know can i have an impact in that space and also make a living out of it?

00:23:21: Kind of didn't click.

00:23:23: so I looked around for one year um... And I think there are opportunities out here but just did not quite feel right.

00:23:32: And so eventually, again I got into contact with RobCo kind of coincidentally and that somehow clicked immediately.

00:23:42: It really puts together your field of knowledge with machine learning also the whole vision topic... ...and then obviously robots?

00:23:52: That is a field where you can move in to their future which is interesting for you!

00:23:58: I think everything worked out.

00:23:59: there is the headquarter here.

00:24:01: It was already at the right size that it's professional, but also still small enough.

00:24:07: so you have access to everybody can know everybody and um And it has just a very interesting and challenging problem to solve and it has Very good opportunities Out there.

00:24:22: i think being in robotics Is You Know The Perfect Time Right Now?

00:24:28: Yeah, I think you're absolutely right.

00:24:29: Robotics is so hot it's really interesting and a lot of things can still be created in robotics.

00:24:36: This leads me to my next question how was that work at Robco?

00:24:41: Not unexpected very fast like one other thing i saw is implemented like you come in and then we know within a week, something goes from idea to being actually part of the product.

00:24:59: Obviously with additional challenge that contrasts to like a SaaS company completely software oriented company.

00:25:08: You also have to move atoms.

00:25:11: So obviously there's huge difference And Robco has this philosophy that we want to have everything under one roof.

00:25:23: Um, so we are creating the hardware.

00:25:25: where you're creating the software?

00:25:27: We have a solutions team and um, we have a team that works with customers from the get go.

00:25:33: And then uh, we don't rely on external partners.

00:25:37: We do it all in house.

00:25:39: So I think this is just an extremely good way To learn every day right?

00:25:46: i can always sit A meeting every week where we talk about challenging customer problems.

00:25:52: I can go out to a customer site, you know when i want To just look at what's the actual problem Out there.

00:25:59: that sets RobCo apart?

00:26:01: Yeah What is really cool About RobCo as well which i don't think many people Know?

00:26:05: The headquarters in the center of Munich But also the production facility Where robots are created and produced.

00:26:12: The actual hardware is produced Is In That Office.

00:26:16: It's basically, you walk from the coffee machine area.

00:26:21: You walk past some desks and then directly with engineers who are actually building robots for their customers and testing them.

00:26:30: I think that is very unique at this point in time.

00:26:32: For sure That your so close to all departments together And it also a success formula To achieve velocity of creating products.

00:26:43: Yeah, definitely one thing which makes it really attractive.

00:26:47: You just the work environment.

00:26:49: we don't have too much space.

00:26:50: I have to say you know this is what other things that come with being in the city center but where right next to TUMunic were in a very vibrant street where he can choose among fifteen different places to eat and When if you're in the office, as you say I mean.

00:27:11: You just walk ten meters and your next to the production guy?

00:27:16: And i've seen all the photos on your whiteboard.

00:27:22: who's working at Robcoe .And it was there a year ago but now half of them are

00:27:30: full.

00:27:31: So how international is your team right now ?

00:27:37: So, Robko was started as a spin-off from the Technical University.

00:27:44: The students that came from there are local and who founded the company were... Meanwhile I would say ninety percent of the resumes i see is international.

00:27:59: When

00:28:00: you're

00:28:00: looking for people we just sitting here in this podcast format obviously but if like encourage some people to send in their applications.

00:28:08: Who would that be and who would you invite?

00:28:10: To join Robco,

00:28:12: I mean we're right now where were hiring into product development team on all fronts from controls hardware software.

00:28:24: but obviously when we think about the innovation side We are building this robot learning based a team.

00:28:35: It's both on the software and hardware side.

00:28:37: So we're evolving, how you interact with a robot but also how it looks like, what sensors might have... And so we are kind of building this AI first-robot system.

00:28:54: Amazing!

00:28:56: Is that something where applicants could join RobCo?

00:29:01: can expect that they will be working at something, changing and sculpting the future.

00:29:07: And it's really a forefront of technology?

00:29:10: Absolutely!

00:29:11: An AI first robot is not something you see in market right now.

00:29:16: There are classical industry robotics which has slightly different focus but very well established for many decades.

00:29:25: But it's mostly, you can see as a robot that is completely blind and just repeats something then was programmed for million times of day.

00:29:35: And on the other hand we have humanoid space which really cutting edge research in many areas solve slightly wrong problem.

00:29:49: so what were trying to take these two things?

00:29:52: put them together come up with some thing that is really, really useful and where a customer of ours might say hey wow this.

00:30:01: This is something I'm willing to pay for because it's a real painkiller for my business

00:30:05: problem.".

00:30:06: Yes!

00:30:07: Really interesting...I think that RobCo is at very an interesting space right now in time as well.

00:30:16: And tell me more about your role?

00:30:18: It's called The Principal Engineer.

00:30:21: bit more about like, how does your day look?

00:30:23: Like what is you're... You know.

00:30:25: What do you do?

00:30:26: and basically um..what's role in the company?

00:30:31: So it's I think two-fold.

00:30:32: one is that i'm looking a little bit across the whole product engineering team And make sure That we are building The software mostly In the right way.

00:30:45: Obviously im coming with these twenty years of experience.

00:30:49: I've seen a few things here and there.

00:30:51: Uh, may-I want to make sure that we have a defined set of things that we wanna build at.

00:30:56: the communication inside of the team is done the right way?

00:31:00: That things are written down in discussed and decided the right weight.

00:31:04: so let's one thing And then the other thing is said.

00:31:08: i bootstrapped uh what would be called The Autonomy Team which is our like first venture into an innovation team after the company was founded, right?

00:31:24: So when you found a company.

00:31:25: You have innovation came out of university and then make it to business and scale up their businesses.

00:31:30: so that question is what's next?

00:31:33: And we're looking at how can build this generation without going completely off tracks with pie in the sky, R&D project.

00:31:45: So kind of balancing that building their team, building the culture and then eventually merging that into the whole product development organization.

00:31:54: And that's a challenge I'm dealing with all time

00:31:56: right now.

00:31:57: very interesting.

00:31:57: so you're basically looking at all the departments being in touch with everybody?

00:32:02: And so really understanding where we can push they new products.

00:32:07: when were talking about communication what is the main communication language as Robco?

00:32:11: It's all English.

00:32:14: So I took that from a former manager of mine who was German, but lived in the States a lot.

00:32:19: he even doing one-on-ones and English Even if we were just talking simply because you said You know?

00:32:24: Yeah Just have to have this muscle memory

00:32:26: makes sense

00:32:28: And make sure that you don't switch all the time.

00:32:30: i think That is really bad sign Because always exclude somebody Who might not be as perfect in german.

00:32:37: No it makes absolute sense.

00:32:38: then obviously The words.

00:32:39: Sometimes when you've been talking in German too much, words will not come up in your mind.

00:32:45: In English and that will hinder the conversation where as the other way around it'll always be in the english mode And is easier to communicate.

00:32:52: but this makes they make Robco really inclusive as well.

00:32:54: right if You're coming in from I don't know some Other fields Some other countries some other language and you are good at English again directly with Robco and start communicating inside and just start.

00:33:11: Yes, absolutely!

00:33:12: That's really cool.

00:33:13: I like that.

00:33:14: thank you for answering my questions because i think it is interesting to understand who you are but also how RobCo works.

00:33:24: the company actually functions what kind of team members they're invited too join RobCo?

00:33:35: And they're also really interesting because.

00:33:40: They are like just fast questions, some fun ones and you can just answer them within ten fifteen seconds.

00:33:46: one or two sentences.

00:33:47: something that all right.

00:33:49: coffee your energy drinks hard.

00:33:51: how many per day?

00:33:52: Absolutely coffee at least twice a day.

00:33:56: I don't start the without good cappuccino and usually in the end of along walk.

00:34:01: where do get it could come from as an extra question but i just thought off.

00:34:06: Last thing, last one I bought in Garmisch.

00:34:11: There's a local roster there which i can recommend.

00:34:13: Okay

00:34:14: very good Very Good Early bird engineer or late night problem solver?

00:34:19: What is your...

00:34:23: It used to be Late Night Problem Solver.

00:34:25: Now when you do that You cant sleep.

00:34:27: So I shifted it more like the early birds Or kind of In the middle.

00:34:32: Also good

00:34:35: In the

00:34:36: transition.

00:34:36: Yeah, late to early I

00:34:38: have to say robotics engineers are surprisingly often early birds.

00:34:41: interesting

00:34:42: i don't know why?

00:34:43: Interesting slack messages or walking over to someone's desk.

00:34:49: both is fine.

00:34:51: Both has different pros and cons

00:34:53: interesting.

00:34:54: yeah first thought on your very first day at Robcoe.

00:34:59: oh i'd have to remember great team.

00:35:06: yep That's cool.

00:35:08: Office, remote or hybrid?

00:35:10: What actually works best?

00:35:14: I worked remotely for five years.

00:35:17: that works great especially when everybody is remote.

00:35:22: i think what doesn't work if somebody is remote and the others are in an office?

00:35:27: yeah

00:35:28: it's also the exclusion point you mentioned.

00:35:31: but how does Robco handle?

00:35:33: obviously the engineers who're building machines.

00:35:35: they have to be on site.

00:35:37: How is it solved for the other employees?

00:35:41: We have people who are remote and hybrid.

00:35:44: It depends on their role, of course when you deal with robots obviously often they need to be onsite.

00:35:54: I would say that i'm definitely much less working from home than used But we're very flexible in that regard.

00:36:04: Like from my role, particularly I'm mostly in the office but you know sometimes one or two days at home is also fine.

00:36:11: Something that i find really cool about RobCo is when I come into the RobCo office... ...I do have this immediate feeling of culture and community And people being there using space to work Also use it as a space for meeting each other not just for meetings, but meeting each other.

00:36:31: So I always have this sense when i walk through the Robco office that there's people gathering talking about stuff.

00:36:38: they're working on changing ideas and also being there enjoying having an office and being inspired by others which is often when you are doing a hybrid or remote setup it really hard to get by.

00:36:52: And the culture really feels, like everybody smiles.

00:36:54: Everybody's happy that people are... We come into the office and nobody knows us right?

00:36:59: Then it is they're hey how you doing everything.

00:37:01: I really liked that feeling.

00:37:03: so this just my take on

00:37:05: what we feel.

00:37:06: to be honest i think many people there would do their job as a hobby as well.

00:37:12: yeah It's such a cool area to work in.

00:37:15: That's good Music while coding productivity booster or distraction

00:37:20: productivity booster, but I prefer actually podcasts or some kind of text.

00:37:26: I don't know why my ADHD brain can deal with that

00:37:29: interesting?

00:37:30: That's very interesting because that would usually be more of a distraction.

00:37:33: right then you have

00:37:35: if music than blues.

00:37:37: Interesting!

00:37:40: How much of your week is real?

00:37:41: engineering versus meetings

00:37:44: shifted alot over the past few weeks and months.

00:37:47: When I bootstrapped the team was mostly engineering, I was very much still in IC until October November.

00:37:54: now it's all about building the team.

00:37:56: so more meetings, more coordination but that's different for other people on the team.

00:38:03: of course.

00:38:03: i think there is plenty time to really do your work

00:38:07: Still.

00:38:09: Average number of browser tabs open right now?

00:38:14: usually about thirty to forty over the course of a day and then I'm trying to trim it down.

00:38:18: Okay, how often does robot or model work on the first try?

00:38:25: Never

00:38:26: okay.

00:38:27: explain why that can never happen

00:38:31: because its atoms is reality And previously you have concept in your head.

00:38:39: It's virtual Its an idea space.

00:38:41: so i think thats why really being where the rubber meets the road, things get real is so important.

00:38:48: Even if you build a robot in your lab environment and put them into factory different kind of challenges.

00:39:00: How closely do software and hardware teams collaborate?

00:39:04: Very closely they sit next to each other.

00:39:08: Are you more in prototyping mode or optimization mode?

00:39:12: The team Like the whole product development team, we have very mature parts of the software which work out-of-the box.

00:39:23: And then what I'm doing right now is very much prototyping like use the scientific method for finding how things work making sure that as efficient way it's possible to try and find out questions you want answer but its definitely prototyping mode.

00:39:46: How much time do engineers realistically get to learn new technologies?

00:39:50: As much as it takes.

00:39:54: We're still relatively focused, like we wouldn't go completely on a sidetrack just because something is cool.

00:40:01: but then within our domain everything's new right there...as I said its an innovation field.

00:40:08: There are no market and proven concepts.

00:40:12: everybody in The industry that is working on it, is trying to figure out.

00:40:18: And that's what we're doing as well.

00:40:19: Well

00:40:21: Who decides on the tech stack?

00:40:23: Engineers or management?

00:40:25: Oh absolutely engineers figured out.

00:40:30: When was the last time you thought oh That's a really smart solution.

00:40:35: I think almost every day like people are smarter than me.

00:40:40: so um i get To see new ideas coming up all the Time.

00:40:43: Amazing failure culture.

00:40:48: Do you have fast experiments or careful validation?

00:40:52: Fast experiments.

00:40:53: I think we need to have the safety, that people can make... Or it's not even a mistake!

00:41:00: A failed experiment is something they learn from.

00:41:03: so what were trying to extract are bits per time unit.

00:41:07: And can engineers push their own ideas beyond the roadmap?

00:41:12: Absolutely i mean there also part of building the road map for most parts.

00:41:19: Yeah, how early are engineers involved in the product decisions?

00:41:24: They're part of the whole conversation all the time.

00:41:28: We have a product management which bridges the gap between customers and product engineering.

00:41:35: There is obviously meanwhile some structure.

00:41:37: then like there's an elites level And you get to discuss these things on that level as well.

00:41:43: But then um You know engineers are involved at every point in the process

00:41:48: And do.

00:41:49: how often the features come directly from engineers ideas?

00:41:52: Or is it like, How's that driven

00:41:54: same thing.

00:41:55: I mean, It's just input from different sides.

00:41:59: Do you see the impact of your work?

00:42:01: quickly or much later

00:42:05: as i said very Quickly.

00:42:06: they turn around time and The company Is extremely short.

00:42:09: nice Who would not be happy working at Robco?

00:42:17: Well If you've grown up in a very corporate culture with long processes and timelines, heavy engineering then Robco is probably not the right place.

00:42:32: We are still at a phase where we're trying to be very nimble And make do with quick decisions.

00:42:43: prototypes really focus on delivery speed.

00:42:47: Yeah, what does someone need to bring?

00:42:50: To really thrive at Robco.

00:42:53: I think Really the energy to get things on the road.

00:42:57: Okay.

00:42:59: and What surprises?

00:43:00: new hire is the most that Robco?

00:43:01: um

00:43:03: what surprises?

00:43:03: people know most i would say The level of competence given that it's such a young team.

00:43:13: Many.

00:43:13: for many people It's the first job after college And if you work there for two years, I mean they have seen all the different pieces of a robotic system.

00:43:26: And I think it's for young people.

00:43:47: It's the best time you can ever choose to go in such a company, because you'll be able to form different kind of connection with people and see so much and learn so much faster would say five times ten times speed than you could like big corporate or something like that where your only allowed to touch your small space.

00:44:07: but yeah is I can absolutely relate to.

00:44:11: Yeah, there's less structure but in a big corporation feels like throwing pebbles at an elephant right?

00:44:18: You're not really able to influence the whole thing and that is different here.

00:44:22: How has performance measured Adropko ?

00:44:26: We have defined HR process where There are three hundred sixty degree review A few times per year But it isn't a lot of ceremony compared big companies that might spend weeks in calibration meetings and so on.

00:44:47: The ceremony is very, very much trimmed down.

00:44:49: nice.

00:44:52: to finish up two more questions.

00:44:53: describe Robco as an employer.

00:44:56: In three words

00:45:01: rebuild the robot market That works.

00:45:08: And last question would you recommend Robco?

00:45:10: To a friend looking for an engineering role?

00:45:14: Of course, I would.

00:45:18: Clemens thank you so much for taking the time and really explaining a bit more about your role in really going deep into all these questions i just asked And then explaining how your path was?

00:45:28: It's interesting to find a matching between RobCo and themselves and understand why RobCo is such a driven company.

00:45:38: Also it's not making robots but creating the future.

00:45:43: You explain that really well and how driven you are to make it happen.

00:45:47: So thank-you so much, everyone for watching our podcast episode!

00:45:52: And obviously if you liked what we just saw please like & subscribe and recommend this podcast also to your friends.

00:46:00: If somebody who is an engineer would like to join the amazing team then definitely let them know that RobCo is hiring.

00:46:07: Maybe they will be next new hire at RobCo.

00:46:11: Thank-you See you next time when we're back in RobCo's

00:46:26: podcast.