This FAQ is a supplement to the SDML-KIP overview document.
No, this is a community project, and the mentors are volunteers. Participants will not pay anything if accepted into the SDML-KIP program. Note that some participants may voluntarily choose to purchase cloud computing resources. This is not required since there are many free resources at Kaggle and elsewhere. Participants may also get recommendations for good books or courses from mentors or other interns. The program is not affiliated with any commercial interests, mentors will not push interns to purchase anything, and interns will never be obligated to purchase anything as part of their participation.
Yes. Cohort 1 is scheduled to run May through July 2024. We hope to continue with multiple cohorts per year.
Unfortunately, we are not prepared to work out the logistics for remote participation for Cohort 1. We will examine remote participants in future cohorts if we feel all of the program goals can be met at a similar level of quality for remote participants as with in-person participants.
Mentors will be heavily involved and very hands-on. Mentors will be part of the Kaggle team and will also be working on the competition. They will attend every Saturday working session (barring an occasional personal conflict), teaching techniques, providing feedback, and suggesting assignments. Participants will also have a private Slack channel where they can ask questions and request help from mentors throughout the week.
Yes, participants will learn many machine learning practical skills, including:
- Exploratory data analysis to understand the data
- Clearly understanding the objective and metric(s)
- Standard ML coding skill and ecosystem
- Manipulating data with packages such as pandas, numpy, and PyTorch
- How to train models, from logistic regression to gradient boosting machines to advanced deep learning models
- How to rigorously validate and improve models
- Effective communication about technical subjects
Participants will also learn tools like how to use Kaggle to pick up tricks from the community.
Since the SDML-KIP program is built around Kaggle competitions, participants are not so likely to learn about commercial tools and ML operations.
In addition to the real-world modeling skills that will be practiced and enhanced during the SDML-KIP program, participants will have concrete work they can link and share afterward. This will include notebooks, code, GitHub contributions, discussion posts, competition results, and Kaggle ranking points. All of these can be leveraged for blog posts or put on your resume and shared with future employers. We hope that interns will also leave the program with more confidence, an ability to learn new things faster, and with additional technical contacts.
Not necessarily. Part of the design of selecting participants for a given cohort will be to choose people at a similar level, maximizing the overlap of things that different interns want and need to learn. It could just be bad luck if a cohort is composed of people at a skill level that is different from yours.