Resources

Generating Research Project Ideas

Finding a research idea is usually the hardest part of the whole process. This guide walks through narrowing your focus, reading what others have done, figuring out what they haven't, and asking whether you can do something about it.

How to Design Experiments

Most students arrive at experimental design backwards. This guide is about developing a clear account of what each experiment is supposed to show before the experiments start, which saves time and produces research that is easier to write up and defend.

How to Find ML Research Opportunities

This guide covers how to find research opportunities, not how to pick a research topic or how to write up your results. SAIRC has separate resources for both of those.

How to Get Free Compute

If you're doing AI research as a student, you're going to need compute. Between notebook environments, cloud credits, and serverless platforms, you can get pretty far without spending a dollar. This page collects the best options we know of.

How to Learn PyTorch

PyTorch is the framework that most ML researchers write their code in. If you want to do ML research, reproduce published results, or read someone else's training script and understand what every line does, you need to know this framework.

ML Forums & Communities

A lot of how the field actually moves happens in conversations that never make it into papers. Understanding what people actually think about a topic usually means spending time in the communities where those conversations happen.

How to Learn Machine Learning

Machine learning sits at the intersection of computer science and mathematics. The CS side gives you the tools to build things; the math side gives you the language to understand what those tools are actually doing.

Machine Learning Opportunities for Students

Competitions, hackathons, and challenge programs are all ways to get experience where the outcome is uncertain. This resource lays out the main options available to students, with an honest sense of what each one is and is not good for.

Beginner & Intermediate ML Projects

The fastest path from 'I know some Python' to genuine ML competence is building things. The twelve projects in this guide span tabular data, computer vision, NLP, time series, and generative modeling.

How to Publish Machine Learning Research

You finished a research project. Now what? This guide covers the main avenues for getting your ML research out into the world. The right choice depends on what kind of work you did, how polished it is, and what you want out of the process.

Summer Programs in Machine Learning

Summer is one of the best windows a student has for serious work. For students interested in AI research, most of these programs offer real mentorship and a chance to see what research actually looks like before committing to it in college.

How to Write a Machine Learning Paper

Most student researchers treat writing as something you do after the research is done, but that instinct is backwards. Writing is the process that actually sharpens your ideas into something defensible.

Open-Source AI/ML Projects to Contribute To

Contributing to open-source ML projects builds real skills and gets your name into codebases people actually use. This guide covers finding appropriate issues, navigating the pull request workflow, and a curated list of active projects where student contributions are realistic.

How to Properly Cold-Email Professors

Cold-emailing a professor is one of the most accessible paths to research mentorship, and one of the most commonly mishandled. Professors who mentor high school students care about demonstrated preparation and specific interest in their lab's work, not credentials. This guide is about getting that email right.

Introduction to Subfields: Reinforcement Learning

Republished with permission by Devansh, writer of AI Made Simple. RL is one of the big 3 paradigms of ML Research (along with supervised and unsupervised learning). RL is based on teaching a model to maximize a reward function.

Research Grade Code

Research-level code is quite different from traditional code in that it has to be reproducible, rigorous, and accurate. There's a much higher bar and it requires careful attention to detail, which is what this guide seeks to teach.

How to Keep a Research Logbook

A logbook is an important part of the research process. Three things depend on it that are easy to take for granted until they go wrong.

Interactively Learning Coding And AI

Typically, people learn to code via tutorials and informational videos before they try to write code themselves, but they find that they retain almost nothing and certainly don't understand the content. Passive reading and passive watching produce familiarity with concepts, but unfortunately, not the ability to apply them. The platforms in this guide require you to write code rather than read it and return immediate feedback when something breaks.

ISEF & STS Research Competitions

Science fairs represent the highest level of high schoolers' original research due to its strict evaluation and rigor. As such, they're taken extremely seriously, not only due to the trophies and scholarships that can be won, but also the fact that doctors and other field experts are critiquing your methodology and experimental design.

Introduction to Subfields: Mechanistic Interpretability

It seems like a lot of AI research right now focuses on making models better at tasks. Instead, mechanistic interpretability asks what the model is actually doing when it performs a task? This is significant because a model can achieve high benchmark performance while doing something completely different from what you think. As such, interpretability methods can tell you entirely different stories than behavioral evaluations alone.