I am also an EE and have been able to apply ML to my field, wireless comms. It turns out people skilled in the intersection of two fields are very rare indeed.
I'd suggest you look for an opportunity to apply ML/CV/AI in your industry (deep learning for PCB inspection maybe?). Show the possibilities, get some research funding, do a pilot program or similar. Lead and drag your company (kicking and screaming if need be) into the 21st century.
Then you will have ML/AI on your resume, and recruiters will come looking for you.
Wireline EE here. ML in the context of wireline communications has been widely studied (grant money works that way). However, at least in wireline communications there isn’t a lot of gain to be had. This is because we have good physical models of underlying impairments and our ASICS are optimized using those physical models, so ML isn’t able to improve upon that much. Where it can help is where we don’t have good data on underlying physical parameters of the channel. So in this case, you can get a bit more capacity than you would otherwise. So for wireless where you have signal fading and huge variation in multi path interference, maybe there would be benefit as these phenomena are hard to model.
Also, you may not realize but we’ve been using ML techniques for decades in communications. Gradient descent is used to optimize equalizers; maximum likelihood estimation(and equalizer optimization) used for phase estimation in high order QAM. Plenty of other examples. So you probably are already familiar with much of the basic tool kit. I had a wannabe startup founder in ML tell me that there’s no way I could possibly understand the stuff if I didn’t have PhD in that area in CS (I am physics). I just smiled and nodded.
Thanks for the response.Wireless comms seems like ripe for ML enhancement. PCB inspection is something that has crossed my mind indeed but it is quiet involved due to the sheer number of features ( traces, components, vias, connectors) none of which has pretrained models. Having said that Automation esp FAI( First article inspection) is quiet possible as demonstarted by landing.ai(Andrew NGs) company.Being employee at a publicly trading company, not sure how I can secure funding. Perhaps my focus should be to apply ML directly to something that I do day-day at work.
Would you like to write an auto router for PCB design using ML? A smart one that can indentify components, their properties and choose right connections between them? Have a fast signal going directly and some shitty LED with 10 vias. Or placing power parts with wide tracks and using thin ones for digital signals.
Auto routers already exist and I always route my own PCBs. I am pretty sure the tool vendors are considering or already building up AI expertise to level up their Auto routers. I keep hearing at least in Altium they have come a long way. Gotto try it one of these days.
I'd suggest you look for an opportunity to apply ML/CV/AI in your industry (deep learning for PCB inspection maybe?). Show the possibilities, get some research funding, do a pilot program or similar. Lead and drag your company (kicking and screaming if need be) into the 21st century.
Then you will have ML/AI on your resume, and recruiters will come looking for you.