How I interviewed Blue Origin spacecraft FPGA designer
One day I decided to reply to messages on LinkedIn from a recruiter from the Bezos company Blue Origin, which develops and launches spacecraft. Then I went through two rounds of interviews: a screening and a 6-hour marathon with a dozen engineers and managers. I cannot tell the questions or describe the dialogues with the interviewers, because I signed the paper that I will not do this, but I will show a couple of things that were discussed even before the paper was signed, and also give advice to those who want to repeat this (advice by the way can be applicable not only to the device to Bezos, but also to Mask or Rogozin).
Simultaneously with Blue Origin, I interviewed another company, the interview with which was the same in time – an hour of screening with an engineering manager and six hours of a detailed interview – but the content was completely different. In general, usually in electronic companies, such an interview is just a marathon of tasks on microarchitecture (all sorts of perverted FIFOs, arbiters, caches, credit counters), tricky questions about timing (it comes to time stealing / time borrowing), the intersection of clock domains, a dozen ways to optimize dynamic power consumption when designing at the level of register transfers, etc.
But in Bezos’s company, it was different. First, it was necessary to write an essay on the topic “What do I think about spaceships plowing the vastness of the Universe.” I wrote the following:
A written summary regarding my interests and thoughts regarding Human Access to Space and Space Flight in general
The solar system has a wide variety of worlds that are very different from our planet, Earth. Some of these worlds have liquid water and conditions suitable for forming complex molecules, possibly even life. These worlds include:
Jupiter’s moon Europa and Saturn’s moon Enceladus. Both have liquid water and internal heat. Saturn’s moon Titan has ammonia-rich liquid water, a nitrogen atmosphere, and seas of methane and ethane.
While Mars these days is more dead than the dry valleys of central Antarctica, it may have the conditions suitable for life in the past. Exploring these worlds will definitely add to the science. The pursuit of science is one of the essential activities of humankind since ever. We may even find something practically useful in unlikely places.
For example, the modern technology of PCR (Polymerase Chain Reaction) became a mainstream biotech tool only after scientists found the bacterium, Thermus aquaticus, living in the hot springs of Yellowstone. Before discovering this bacteria, scientists did not believe that life could survive at temperatures of 170 ° F. They now use an enzyme from this bacteria, a heat-stable DNA polymerase, to multiply tiny amounts of DNA to test for viruses (including COVID-19) and create genetically engineered strains of bacteria to manufacture useful chemicals. Biotech is a multi-billion dollar industry, and PCR is a workhorse of it.
Who knows – maybe researchers can find non-DNA-RNA-protein-based life forms or some unusual molecules in Titan and use them for something we cannot even think about right now.
There is, of course, the argument that the planets can be surveyed by robots. Some people even talk about autonomous robots with artificial intelligence. I can see a problem with this argument: modern AI is just a combination of matrix multipliers and finite state machines. No amount of complicated software code, clock frequency, memory size, and creative digital hardware design based on gates and flip-flops can change it. Human intelligence, on the other hand, may work on different physical principles, not discovered at the moment. In particular, science has no idea about the nature of self-awareness, and so far, nobody has observed self-awareness of a CPU, ASIC, or FPGA.
This makes the case of human space exploration convincing. Humans can go to Titan, Europa, and Enceladus, get an idea of what is going on there, and find something robots can miss.
At the beginning of the interview, they had to give a presentation to their engineering group and then answer questions – technical and non-technical. I was surprised that the technical questions were quite basic: about sequential logic, bus protocols – no microarchitectural complications, as in companies that make CPUs, GPUs and network chips.
Of course, space companies, as you can read @amartology, have a ton of interesting specifics. Special methods of writing on the verilogue of finite state machines so that the cosmic particle that spoils one bit of state does not abort the entire mission. Troy blocks, an incredible mixture of sensors, microcontrollers and FPGAs, is a paradise for system designers.
But from the point of view of purely FPGA technology, you could answer half of the questions in the interview in Blue Origin if you just went through the Skolkovo School of Synthesis of Digital Circuits, carefully completing all the tasks, and would also receive a free FPGA board from Alexander Bilenko and practice with her at home.
True, in addition to FPGAs, there were many questions about behavior / teamwork and dedication to the ideas of conquering space. I didn’t impress them on them, they called me a week later and said that they had chosen a candidate who was better than align with the position. I breathed a sigh of relief, since I was going to go to another company anyway (which did not ask humanitarian questions, but instead asked a series of microarchitectural puzzles). At least the space comrades themselves removed the pangs of conscience from me that I wasted 7 hours of the time of important people.
Now about Skolkovo School of Synthesis of Digital Circuits… Last Saturday there was exactly a lesson on sequential logic, about which there will be questions in any space company that you want to go to FPGA (at least to Mask, at least to Bezos, at least to Rogozin). Here is the complete entry:
This Saturday, November 24, the lesson will not be in Skolkovo, but in MIREA (this is next to the metro station on Yugo-Zapadnaya). Here is an instruction from Alexander Bilenko on how to participate in this:
The next lesson will take place on November 27 at the Altair Technopark MIREA (Vernadsky Ave., 86, bldg. 2, Metro Yugo-Zapadnaya)
Topic of the lesson: Analysis of the educational project: recognition and generation of sounds and melodies.
For this lesson, you will need an FPGA board, as well as a Digilent amplifier and microphone connected to the board, and headphones with a round 3.5 mm jack.
We will give all this to you in class, if, for example, you do not have a fee.
By the way, those of you who want to continue mastering the topic at home on your laptop can get a board in class and take it with you.
The lesson will prepare about 15-20 computers with installed software and connected boards.
But if you have a laptop with installed software for classes, it is better to take it with you, in case there are not enough jobs with computers for everyone.
Since there is an access control in MIREA, we must submit a list of students, for this we need to get from those who will be at the lesson on November 27 SURNAME, NAME, PATTERNAME.
Send this information to firstname.lastname@example.org if you intend to attend class.
At the end of the lesson, you will be offered several tasks on the topic and everyone who successfully completes them will receive a Yamaha YES-23D Recorder as a gift, on which you will be taught to play a simple melody by the flute teacher Maria Belichenko.
We’ll show you how to design a circuit board that recognizes the tunes you play and even sing along.
So, let’s start forming the group for November 27!
The next lesson will be like on ChipEXPO, where circuits on FPGA boards recognized the melodies played by Maria Belichenko:
But this time, everything will be more interesting, because in addition to FPGA boards, you will personally receive each recorder, and you will play melodies for FPGA boards on them, like Indian fakirs play melodies for cobras. I don’t know how many places there will be, but since many people are too lazy to go to South-West, some places should remain.
There will also be field sessions at the HSE MIEM and at MIET in Zelenograd