ML/ AI advice for a Senior College Student

Question

Hi All,I hope everyone is having a great holiday. I am an undergrad student with one year left until graduation; I know a few languages and don't have any experience in the front end or back end. I know python and took a beginner ML course.  I want to get into ML/AI, but I don't know the right path to learn. I don't know what field I want to be in. I am using Codecademy to learn ML now. Is this a good resource? Are there better resources? What can I do in the one year I have left to break into the field of ML/ AI? Do I need to learn full-stack to be an ML AI engineer? Are ML/AI engineers considered data scientists? are they SWE? Thanks

undertone · Accepted Answer

I feel that increasingly ML/AI is breaking out into its own role outside of just SWE. And the skills are more specific to the role than to general software. More so than anything I’d work on your own projects. Try to build and fine tune your own models, and learn some of the fundamentals behind using datasets to train. Even just understanding simple logistic regression inside out will put you so much farther ahead. Just building a project and putting it out there for people to use will also inevitably develop your full stack skills as you’ll have to create something end to end. Find a friend (ideally a bit smarter and knowledgeable than you) and just work on something.

Guantanamayna · Answer

What is your degree in? One year left but you don’t know any front or back end seems…odd. ML/AI is highly specialized and getting something like that right out of school will be tough without internships already showing that background. It’s an entire field and specialty with their own specific requirements to learn. Quite different from full stack SWE, with some overlap with data science. I would recommend looking at job postings to understand what you will need to learn in order to get hired.

Tridom · Answer

Try for an internship, or an independent study with your ML professor. If you go for the latter, do the work with the intent to publish. It’s a ton of work, but a publication can be a springboard into the industry, and the professor you work with will have their own motivations for getting publications with their name on it. Otherwise, you’ll be working your own projects. I’d recommend Introduction to Statistical Learning as a resource for honing your statistical basis, which is typically what I’d expect to see a CS BS grad could improve.

PaulDeLuco · Answer

Big Data and AI/ML are negatively correlated with respect to a vendor's ability to sell their respective offering. This means ML/AI (from a sales perspective) is a hedge for when a big data offering failed to sell.

If you want to work in ML/AI, then understand that your career highs are mostly possible when customers lack data and infrastructure (i.e. small to medium size companies). So, the majority of your career is going to be doing ML while also setting up infrastructure or developing a data source, usually in an informal start-up/R&D setting by another name (i.e. product development).

Most people hate cleaning data. It is a thankless job. It is also a low impact job, in the sense that it helps someone ultimately sell a data product, but it won't give you much career progression or learnings.

When you add informal start-up conditions into the mix, you should expect a lack of protection by your managers. If they don't have a data or ML background, they will likely not be able to explain your work nor advise you and you'll quickly find yourself face to face with senior management who don't care about details or R&D processes. This scenario may seem like career progression from the outside (you're talking with your COO or CEO), but it's equivalent to an IT -hero culture that produces silos.

Setting up infrastructure is more interesting and progresses your career toward site-reliability, and IT/operations, but chances are you won't be doing ML/AI and instead be more logging / telemetry / data-engineering focused.

I would suggest deferring ML/AI aspirations and instead position yourself to work in or on the peripheral of big data first. (i.e. IoT development for sensors which could be feeding an MQTT broker to then feed a downstream kafka queue which would ultimately feed a big data product, or instead learn about data meshes and big-data streaming, or just learn a cloud platform, etc.)

ML/ AI advice for a Senior College Student

About

Software Engineer

Members