Advice for career development

ML is just hype , jobs are also lesser than sde

From someone who works in one of the few companies that do ML at scale: be vary of the hype. A few thoughts: - Uber is one of the few places where we employ a lot of data scientists for machine learning. This is because we have both the amount of data, as well as real-world uses cases (fraud, routing) etc - There is a huge amount of ML-talent overflow already. For every ML job we get 10-50x qualified candidates. If you don’t have 2-3 YOE or a phd is something practical, you don’t even get to an interview - The actual work is mundane. Everyone coming in expects to work on sexy models that change the world. Then you end up working on outage calculations, repetitive experiments... the novel stuff is already taken by someone with 5 YOE already (remember how we hired a lot of smart people who want to work on this) - Because of mundane work, you see huge turnover with ML talent. Data scientists / ML engineers move teams and leave the company (and every other company) very quickly, once they hear of a new and interesting opportunity. So any job that is not filled by a referral or hiring someone with 4-5 years industry experience is pretty mundane - on top of getting this large inbound if engineers - Product launches with simple logic (if... then... else....) regularly have better results than ML models. There are few areas where ML brings benefits at scale... and at those places, we still have 2x as many software engineers build the system around it (visualisation, analytics etc) - I see every 2nd software dev applicant wanting to specialise in ML and a bunch of them coming with nano degrees. Then surprised there’s no ML openings If you want to work on interesting things, at scale, software engineering is still a much easier path to enter. With ML, forget that you’ll be able to get interesting jobs, unless you have a PhD from a top ML lab. It will leave you competing with masses of people thinking the same way, in the end probably ending up at a startup that has neither the data quantity, nor the expertise at this. But at least you’ll have ML on your job description and it will compensate for you making half that of the backend dev sitting next to you. Specialising in ML and then only shooting for ML jobs is the same career path as specialising in game development and then only applying to these jobs IMO. And I’m someone who’s done an ML masters 10 years ago, before it was cool :)

Trust me you'll end up modifying yaml files to tweak some config. I spoke to an HR person recently, they said there's no shortage of AI/ML personnel anymore

Distributed system is the way to go