01
Machine learning
How a model learns relationships from examples. Separate training data from test data before measuring performance.
Train a baseline
Stuyvesant High School · New York City
Surf the wave.
Make an impact.
A student-led AI club where research becomes publications, ideas become working projects, and learning reaches beyond the classroom.
Curiosity, connected.SurfAI / Research
First place · NYC TerraFairSoftware / Robotics category
1stTerraFair · Software / Robotics
7Published journal issues
4Hands-on learning labs
15,010Site views · archived total
Learning alongside
our collaborators
Research.Ask better questions.
Share useful answers.
Share useful answers.
Read. Research. Publish.
Build something worth sharing.
Research and publications are at the center of SurfAI. We turn curiosity into questions we can test, explain, and share.
A publication-heavy club
Our journal makes complex developments in AI, finance, and science easier to investigate and discuss.
Experience that goes beyond a demo
We bring hands-on research and publication experience, including first place in TerraFair’s Software / Robotics category.
AI education that helps people
We connect technical curiosity with community needs, especially healthcare and medicine.
A foundation for every curious mind
Understand the ideas behind AI.
Nine starting points. Real concepts, useful examples, and a next step for each.
02
Neural networks
Layers transform inputs into useful representations. See how weights and a loss function guide learning.
Explore the layers03
Language & transformers
Tokens and attention help models relate one part of a sequence to another. Context is useful; it is not a guarantee of truth.
Try attention04
Computer vision
Learn how models locate, classify, and track patterns in images. Lighting and camera angle can change a prediction.
Open the camera lab05
Generative models
Diffusion and autoregressive models generate new samples. Compare a plausible output with evidence from the real world.
Understand generation06
Reinforcement learning
An agent learns from rewards. A high reward can still miss the goal if the reward is poorly specified.
Study reasoning07
Retrieval & agents
Retrieval adds relevant sources; agents can use tools. Build in checkpoints, permissions, and ways to recover from failure.
Trace a workflow08
Evaluation & responsible AI
Test beyond a single accuracy score. Study precision, recall, bias, privacy, and the cost of a wrong answer.
Run an experiment09
AI for health & science
Explore medical imaging and biological models. Ask what was validated, on which population, and under what conditions.
Read the researchThe SurfAI Journal
A record of curiosity.
A habit of publishing.
Seven issues explore AI’s changing relationship with research, technology, finance, and society. Meet the ideas, follow the sources, and bring your own perspective.
Learn by doing
Make the model
less mysterious.
Try live gesture recognition, explore heat and health, examine pet-food labels, or experiment with image resolution. Each lab shows its method and limitations.
