Future of ML/deep learning career

Oracle 🧠failure
May 3 29 Comments

I'm currently working as an IC. Got a good offer as applied ML engineer. I'm working on distributed systems right now.
In my future job I'll be working as applied ML engineer productizing ML research work produced by research team.
How do people on blind see the future of this kind of work. Do you guys this it just as a hype?

Current TC 130k
Future TC 200k

comments

Want to comment? LOG IN or SIGN UP
TOP 29 Comments
  • Uber lost a $B
    For the people who do research, no. For you, yes.
    May 3 10
    • New / Eng abuhr3i
      You don't want to be too long in the position of productionizing other people's research. Only do that long enough to learn and get exposure. Then use that learning to do something innovative. Those lazy ass researchers can go F themselves or perhaps learn how to write production code.
      May 3
    • Uber lost a $B
      ML engineer is too vague. It is their actual expertise that matters. Distributed system? Low-latency numerical optimization? Most are good and desirable, just not the guys who write adhoc training script calling people's API.
      May 3
    • New / Eng abuhr3i
      What you described is data scientist. ML eng builds the system and the APIs other people use. Obviously these terms are misused and sometines overlapping. The best distributed systems expert in my team has data scientist title. Titles don't matter. Your skills matter.
      May 3
    • Uber lost a $B
      I didn't say anything about title. OP should know best what responsibility his job offer entails. Unless you are at the few places with the scope, scale and business need to justify you writing ML system at scale, you are very much calling other people's API regardless of what you wrap your code in or what your title is.
      May 3
    • New / Eng abuhr3i
      Writing ML systems at scale is practiced in many companies, more than you think. At some point you call some APIs but how to build your systems around those calls requires ML expertise, cause they're never sufficient.
      May 3
  • Uber / Eng Trigger
    u see how testing jobs became obsolete and became part of development ? as the ML research gets more diluted with more researchers, ML eng will become part of research and will lose value
    May 3 8
    • New / Eng abuhr3i
      OP you're good to go. Source: I'm ML eng with PhD in ML.
      May 3
    • Oracle 🧠failure
      OP
      Thanks. What are my chances of transitioning to ML researcher from ML engineer after working for a while?
      May 3
    • Uber / Eng Trigger
      I have seen SDETs who r not as good as SWE and could not convert to SWE. They had to go find meager opportunities outside or worse lose job or take a huge hit to salary. See the same in ML eng. They won't be as good as researchers. But may be have to change back to SWE. This is just my opinion. No one knows what's in the future
      May 3
    • New / Eng abuhr3i
      It depends on your experiences at the job to be more on implementation/scaling or on algorithm development and applied research. Why do you want to transition to research? Research is overrated. The ROI and impact is lower. You can't believe how many researchers from top labs in industry come for interviews with my team (engineering) cause they want to have potentially more impact by being closer to the products.
      May 3
    • Snapchat lullerina
      I think ML research is usually seen as being the top of the band in terms of product vision and direction. Or they could do more just publications and research, but using company resource to do so.

      Usually the difference between testers and SWE is that SWE typically had to do a Bachelor’s in CS (although this is a very rough generalization). And for ML research needs PhD and publications. So maybe the difference is in academia education background?
      May 3
  • Snapchat gqkO66
    2nd the opinion that ML engineer is better than ML scientist. As an ML engineer you have (should have, can convince people you have) desirable back up skills. Usually ML scientists don’t have (or it’s difficult to convince people they do have) the suit of SWE skills, eg production tools debugging, PR reviewing, api glueing, leetcoding. Also that ML PhD is going to get you the side eye by recruiters and hiring managers
    May 3 2
    • Snapchat lullerina
      What makes those SWE skills more desirable? It feels like those can be picked up by any ML scientist, but may be harder for SWE to publish papers
      May 3
    • Snapchat bFCN68
      It's the other way around actually. SWE can pick up ML skills and try random shit with approximately similar level of success.
      May 3
  • New / Eng FAANG_
    How research has no future in ML/AI! I think researchers are most valuable than Software engineers in ML/AI field . The problems in ML/AI can be solved only by new researches and trials . Programming is not that much in ML/AI . And of course ML/AI engineers have a good future. I encourage you to go on ML/AI engineer.
    May 3 2
    • Oracle 🧠failure
      OP
      What is the difference between ML engineer and AI engineer?
      Sorry too many terms keep popping up these days and there is no industry standard as which role maps to what responsibilities.
      May 3
    • New / Eng FAANG_
      I am sorrry I will update it . I mean ML/AI . But for your question. AI has many tools and techniques to be achieved , one of them is ML. Also ML has many tools and techniques , one of them is Deep learning . Iam not ML/AI engineer but I read some materials like deep learning.ai . It is good source to learn about ML/AI.
      May 3
  • New / Eng FAANG_
    I got your point of view . It is reasonable. So what is your advice for software engineer if he wants to specialize in something valuable?
    May 3 0
  • Oracle 🧠failure
    OP
    How is ML engineer different from software engineer in machine learning.?
    May 3 0
  • Intel hw2sw!
    +1
    May 3 0

Salary
Comparison

    Real time salary information from verified employees