Apologize if this is cliché since it seems like every new grad wants to work on a "ML" related team. ----- I'm currently wrapping my internship as an applied scientist on an ML team at Microsoft. I'm happy with my role/team and like the work, but since fall recruiting season is coming up, I thought might as well apply for some other full time roles too. But I am having trouble identifying which positions to even apply to. For example, there are some "software engineer" new grad positions at Facebook, PayPal, Asana, and others, but it does not seem geared towards ML type work. More like web development, which I have no background in. I know some smaller companies like Adobe, Groupon, etc hire new grad for ML engineer roles. Non-tech companies (e.g. banks) and startups also hire new grad MLE. What about other companies? I see many companies, like FB, Google, do have ML postings, but geared towards PhD grad (I don't have a PhD, I also don't have significant research experience). Also it seems some of those roles are doing pure ML research, which I have no interest/qualifications for. One thing I really enjoy about MSFT is the ML/data science community. Very easy to learn about what people are doing in other problem spaces. Lots of mentorship opportunities. I think at a smaller company it may be harder to get that exposure. The one dislike I have is with the engineering culture on my team. Very slow-moving in general, even though people also work long hours. People can be resistant to trying new ideas/models outside of the status quo. Transferring teams require an internal interview loop (so might as well look externally too). Most of my prior work has been in NLP. But I have a pretty wide breadth from school (stats background + some courses in ML/NLP/CV/RL/etc to finetune my knowledge), so I keep my options open. Any advice on where to apply would be appreciated :) My profile: BA and MS CS grad from a good school (one of {Stanford, MIT}) Prior SWE and ML internships Have done some research projects, no conference paper tho :( So I can't apply for AI residency programs at FB/Google/etc
Unfortunately, I have no idea as for new grad Applied ML roles at other companies, but it seems like MSFT would provide a pretty nice opportunity for you 🙂 I’m also a MSFT Intern (SWE), and I’m pretty interested in Data & Applied Sciences (DAS) at MSFT. Could you please expand upon your experience this summer? What advice would you have for someone trying to transfer from SWE to DAS?
I think there are many flavors of DS/AS at Microsoft. My experience was mainly focused on taking ideas from papers, past internal projects at MSFT, etc, and implementing to fit my problem space. If you want to go from SWE to DS/AS, it would help to have some machine learning knowledge. Not extremely deep like neural nets (unless you want to interview for those teams). But your knowledge of statistics and classical models should be solid. I would recommend ISLR (book) as a resource. Some concepts are a bit outdated, but otherwise most content is excellent.
Thanks so much! I’m studying applied math & statistics in college, so I think I have a solid probability & statistics background. Would focusing on ML/NNs be the next step? If so, would it be best to do both ML/NN projects & go through ML/NN resources?
I am not sure about other ML areas, but to be considered for a Computer Vision Deep Learning role, you usually need at least descent research experience or research papers or a PhD, or a combination of these things. I almost never saw anyone get a CV related job without these, even at very small companies. I have been working in ML CV for 5 years now.
Hey, while I do have research publications, they're not in good conferences. How much of a difference does that make? Like does it basically make the publications void in the eyes of companies? Ofc, they are peer reviewed, but still.
There will definitely be more weight if you have CVPR, ICCV level papers. Second tier conferences (WACV etc) are okay too for some companies, but third tier people mostly won’t consider. Also the level of your authorship plays a role. If you aren’t a first author, people usually doubt the contributions. But I am talking about research teams, and not engineering or SWE teams.
But there would be SWE positions in the ML orgs, that would be open to Masters grads without much research experience, and it shouldn’t be too hard to get an interview for those roles.
I see. Would the job responsibilities for those SWE positions still be model building, or is it mostly productionizing models built by PhDs?
For SWE it is usually working towards putting models into production. But you could try some engineering roles too, which would be at the intersection of SWE and research roles. Entirely research roles would be hard for your profile I guess, but if you are determined, you could get the engineering ones.
Apply to Expedia, may not pay as well as FAANG but if you don’t have those papers under your belt I’d go there first. It’s a great team and lots of interesting ML/DL projects. Also space to do longer term research projects and publish some papers alongside shorter (1-2month) projects with high impact/visibility. Best place I have worked at.
India
Yesterday
642
Who are these retards asking for dictatorship in India?
India
Yesterday
580
Modi is a legend, will be remembered for centuries to come
Tech Industry
Yesterday
612
Amazon stopped new green card applications
2024 Presidential Election
Yesterday
629
Heartwarming peaceful protests
Tech Industry
Yesterday
1382
The end of Backdoor Roth?!
If you are open to ML SWE roles G is definitely hiring
Thanks for the lead. What does an ML SWE do? (Couldn't find much info online). Is it more about building the models, or just productionizing them? Also, is the application just to apply through the general university grad SWE posting and then at team matching (if I even get that far) I can request an ML SWE role?
Getting data, building data pipelines, feature engineering at scale, training models, evaluating using product defined metrics, working with deployment team to make sure things are working correctly. For Google you would have to request the recruiter to set you up with an ML themed interview, so you get some coding questions (same as the regular interview) and also a few ml theory questions. You can also switch internally later on but that's also not super easy