Ask HN: Transition from Web to Embedded | Hacker News Reader

Ask HN: Transition from Web to Embedded

23 pointslangsa11 comments

I have been doing web development for about 5 years now and am at a point now where I am getting pretty bored. I'm 32 and have a degree in Electrical Engineering from a good engineering school in the states, but it's been a long time since I've used it.

I'm wondering if anyone has transitioned from web dev into doing more lower level firmware type programming for a job. How did this go? Do you have any advice? Was there some kind of curriculum you followed?

I am worried that specializing more will lead to fewer job opportunities, but I find myself gravitating more towards embedded types of projects in my free time and enjoying them much more than I do any kind of work I do during my day job.

View on HN

I kind of moved the other direction and worked with people like you moving from "high level" development to embedded.

The first thing to note is that "Embedded" is a very wide area, a lot of it is not too special and you might continue doing web development as you are used to, maybe with small adjustment of how you load, debug and communicate with your device.

It's the lower layers, mainly device drivers and BSP (Board Support Package) but even some algorithms close to the hardware or special optimizations, where things becomes interesting and requires specialization.

I worked on those lower layers and have found myself more than once debugging "real software" using a logic analyzer or other Electrical Engineering tools to catch a software problem.

How did it go for those transitioning "down" into embedded? usually quite well, you will need to catch up with the technology, learn some new concepts and make a change in the way you think of things, for example memory or multi processing, but it is definitely doable and enjoyable.

I would recommend it: it’s a lot nicer than web dev imho. I work a lot with the Ambiq boards and yeah, it’s a pleasure. Having months of battery life and optimising every cycle is fun and shows what we could have. Most embedded devs I work with now are sloppy and lazy and often forget that cutting $0.01 production cost means something if you deploy in a minimum of 10million devices. And actually not needing to recharge the battery or just using less energy for weeks to months is also a massive boon in many projects. I find that type of dev fun to do. Makes me happy.

I did a similar transition a few years ago. Used to work with higher-level system integration and web, now got a job as an embedded engineer at a company that makes lab equipment. They had a single embedded developer for the last 10 years who did everything himself, from scratch. We are now stuck with a gigantic monolithic code base where you need deep understanding in the details of the entire system, which makes it very hard to change or fix something.

The problem I struggle the most with is that this industry is 20 years behind compared to modern software development. Since embedded is tied to hardware, and these products have a long life cycle, innovation moves very slowly and there is little motivation to change something that works. I am used to automated pipelines, tests, modern debugging tools etc, and I feel extremely limited going back to a very manual workflow. The use of OSS is almost frowned upon.

I would advise to research what tech stack they are using at any possible new position. Make sure there is a team of embedded developers that work together (not individually on different products). I would look into embedded Linux and Zephyr.

I'm trying to do the transition as well but in the opposite direction. I like embedded far more than web development but depending on where you live it's tough to find good jobs with decent pay in the west.

I feel like a large part of embedded is just telling people to stop counting BOM pennies and buy COTS. It ends up cheaper a shocking majority of the time. I've toyed around with the idea of moving up the stack, but I don't know if pushing protobufs around full-time is any less tedious.

Depends. I've done embedded in high margin products like precision measuring equipment and semiconductors and low margin products like automotive or advertising electronics. Pay and software is mor or less the same, what varies is the amount of bureocracy, documentation and requirements engineering.

Getting good pay means you have to find a Fang with high margins on their HW like Apple or Nvidia or go HFT.

I'm not too upset with the pay because I have a nice niche in the FANGoverse. It's mainly the short-sighted muntzing that I find tedious, where tens of millions in development costs are spent to avoid tens of thousands in BOM costs over the product lifetime.

>I have a nice niche in the FANGoverse

Can I ask what that niche is?

Robotics and custom silicon stuff. I used to do consumer electronics, but that was pretty terrible.

seems to me like buffer pushers would be more viable for FANGS. embedded is fun and interesting but for individual contributor / engineer there is not much money in it. might as well open up a shop backing up ssd drives and reinstalling windows for grandmas.

Embedded has fewer job opportunities. It's maybe one-tenth the size of web programming. On the other hand, there are a lot fewer embedded programmers, too.

Can anyone who knows comment on the relative pay between embedded and (non FANG) web programming?

Reply on news.ycombinator.com