Currently working at at&t Labs as a scientist after finishing my PhD, and current work involves networking and 5G. I want to switch to one of the better tech companies and am confused which domain I should be targeting - SWE or DS. While I do have coded a lot for my PhD, I do not have a formal understanding of software engineering in terms of taking courses. My codes are just a random assembly of things I wrote or found on the internet. Similarly, I have publications on reinforcement learning but I wouldn't claim my background to be quite strong in it. What are some of the things I should keep in mind when deciding where to apply? My goal is to reach a better TC with preferably some elements of research. #tech #datascience TLDR- things to factor in when trying to switch from telecom to software or data science Edit 1: YOE < 1 year, TC 205k
Reading your post, Data Science makes more sense. 1. You have a PhD 2. You can code but arenāt a developer 3. This quote: ābetter TC with preferably some elements of researchā, realistically you do not do the latter in SWE. I would say that you have knowledge of an in demand domain space, my suggestion would be to try to get an interview at a mid sized firm to see how you fare. Best case you get an offer, worst case youāll know what you need to work on.
Research work is something I am flexible with, it's the TC and WLB that would be priorities (I feel bad typing this after spending all those days reading papers š ). I did 2 data science boot camps that are meant for STEM PhD students to break into DS. But I didn't feel it was that rigorous in terms of what I read in LinkedIn and other places on data collection, preparation, building pipelines etc. It was more of applying standard libraries on well designed data sets. Any pointers on what would be a good approach for building a DS portfolio, assuming my current work won't be contributing a lot to it?
Iāll be frank and say that Iāve never really been interested in a 0 YOE data scientistās portfolio when interviewing. In reality the vast majority of DS is cleaning messy data, and then extensive feature engineering. Often the easiest part is the model training phase, as long as Iāve done a good job on the first bit, the latter should be painless. Iām more interested in process and the structure around how you approach a problem. All that said I would say going through kaggle data sets and working through the associated problems would give you a āportfolioā but more importantly give you a process and structure to tackle problems.
What was your PhD in?
Communication and Networking - Electrical Engineering
I have a PhD and switched to DS/ML. DM if you want more info.
Today I Learned
12h
461
How many books do you usually read in a year?
Working Parents
14h
1122
Closed now - thank you all
Working Parents
Yesterday
981
What do you think is wrong with a kid who got rejected by 9 colleges?
Tech Industry
3h
620
Women, help me understand why this is inspirational
Tech Industry
Yesterday
2882
Quitting this Slave life
TC and yoe?
Added