Computer engineering and DSP

[barkbomes9]

I'm currently a computer engineer, but I'm very interested in learning more about digital signal processing (DSP). I'm currently interested in taking a graduate course in DSP and in estimation theory, but I know that most DSP engineers are traditionally electrical engineers. Currently, I'm trying to pursue a specialization in FPGA's and embedded systems. There is also a class in neural networks and IoT that interests me. I would like to know what people's experience with DSP is in industry and if they recommend that I pursue it.

[Digital Larry]

I'd recommend that you pursue things that interest you, especially if it seems like there's an industry around them so you can make a living.  I studied EE and the last year of that was heavy with DSP.  But I never really did much with DSP in my legit career.  Opportunities and fate (?) took it in other directions - more microprocessors, board level engineering.  Ultimately I got bored of that and went into project management, tilting more towards software than hardware.

I've been getting back into DSP purely as a hobby for about 10 years now.  I mostly like it because it does not usually require soldering.

DL

[Blackaddr]

I'm currently a computer engineer, but I'm very interested in learning more about digital signal processing (DSP). I'm currently interested in taking a graduate course in DSP and in estimation theory, but I know that most DSP engineers are traditionally electrical engineers. Currently, I'm trying to pursue a specialization in FPGA's and embedded systems. There is also a class in neural networks and IoT that interests me. I would like to know what people's experience with DSP is in industry and if they recommend that I pursue it.

My career path went:

B.Eng Computer Engineering --> 5 Years as Digital ASIC Designer --> 10 Years as FPGA & Electrical Designer --> 3 years SW engineer for accelerated computing (on FPGA/SIMD/GPGPU) --> 2 years supervising a team on ML/AI and HPC.

DSP typically means writing software from the basis of hardware expertise in high performance computing using C/C++/CUDA. Neural networks is the opposite. You don't get involved with the hardware, and you do most your work in Python and use pre-made packages to accelerate training or perform inference (e.g. TensorRT).

If you are an expert in one, you can shift your career to the other. If you are learning both, I suggest pick one and stick to it because they have related end applications but fairly disjoint skill sets and knowledge requirements.

[elseif]

Your career focus will change over time.  Try to work on signal processing applications that interest you and that allow you to creatively get lost in the work. Then it's fun.  That fun can be affected by the industry in which you work.  If you are in a safety critical industry (medical, avionics, automotive, etc.), there may be regulatory requirements that can greatly affect the fun you will have.  The documentation, testing and certification often costs more than the development of the application itself.  This includes FPGAs and hardware.  I guess I'm suggesting that you learn about these aspects of the industry in which you will be developing applications.

[imJonWain]

I think embedded systems and FPGA related stuff is the way to go, especially if you like it.  You can basically get a job anywhere or even remotely if/when you want. 

I went to school for mechanical engineering but have worked as a electrical/electro-mechanical engineer, which I love.  While start ups and small companies love my mixed background big companies dislike my lack of an EE degree or masters seemingly regardless of experience.  Just my experience.