Mujin Computer Vision Engineer reviews

- I really wish I had more time to allocate to do more experiments, research deeper and fundamental solutions, explore more recent technologies and learn more from the massive data we have access to. But, I think this is normal for a company expanding as fast as mujin - living in japan can be difficult at times. Please make sure to reach out to GA if you need any help.

I feel Mujin’s technology attitude needs to be more open to embrace new technologies. Once our best dev did a decent rewrite of a core package in a lately popular language, to achieve less compile time and better clarity, but our CTO took his achievement lightly, and commented in public about his work as “not super bad, but usable”. Also, I think the company shows much bias against Machine Learning (ML). I understand it’s a marketing strategy to say Motion Planning (MP) has significant advantages over ML that many competitors/robotics companies adopt. In this way, it makes the market see it as an apparently different type of products/solutions. However, MP is just a general methodology, and is also widely used by whoever manipulates robots. So actually everybody is doing some kind of MP. Another reason is that our tech people do not take ML seriously, and do not try to study it. Influenced (misled?) by our CTO, people think that ML methods simply play with statistics and random numbers, and hence are nondeterministic and unreliable in production. They ignore that ML can provide good predictions and guidance to solve complex problems efficiently, and a model can be refined by correction and validation (as done in several robotics companies I reached out to). Results are definitely reproducible if the random seed is fixed. Meanwhile they totally accept that their MP also uses random numbers to sample points, so randomness should really not be their concern. The real reason they ditch ML is that our CTO needs full control in the products, and simply cannot control technology directions he cannot understand, even if it’s more advanced and even if it offers better and more efficient solutions. I think Mujin should take the competitor companies more seriously. When talking about them, our CEO always has a condescending tone, calling them “copy cats” and “inferior robots”, saying Mujin’s solution is the best and most advanced among all. However, I don’t see him hiring business specialists in the back office who can do business research, like SWOT analysis or marketing research in markets in China and America. Top management people seem to think naively they can take some market shares in a country and then just jump right in and do it. As developers, we honestly don’t know what are our strengths and what are weaknesses compared to competitors. We feel a bit blindsided.