The AI Universe

Computers, code, and the ideas every other track builds on.

Algebra, calculus, probability, and linear algebra for machine learning.

What artificial intelligence is, its history, and how to think about it.

How machines learn patterns from data to make predictions.

Neural networks, backpropagation, and the architectures behind modern AI.

Attention, tokens, and the architecture powering today's language models.

How LLMs are trained, generate text, and are controlled at inference.

Designing, structuring, and evaluating prompts for reliable outputs.

Turning text and other data into vectors that capture meaning.

Storing and searching embeddings with approximate nearest neighbors.

Grounding language models in retrieved documents and data.

Adapting pretrained models with supervised and preference methods.

Models that generate text, images, audio, video, and more.

Models that plan, use tools, and act in loops toward goals.

An open protocol connecting AI applications to tools and data.

Orchestrating deterministic and AI steps into reliable pipelines.

Persistent, self-directed agents and their limits and safeguards.

Code generation, review, and agentic software development.

Serving, scaling, and running models in production.

Accelerators, memory, and the hardware that makes AI run.

Deploying and scaling AI on cloud platforms.

Tracking, deploying, monitoring, and governing models in production.

Measuring quality, faithfulness, and safety of AI systems.

Attacks, defenses, alignment, and responsible deployment.

Teaching machines to interpret images and video.

Understanding and generating human language.

Speech recognition, synthesis, and audio understanding.

Learning to act from rewards, from Q-learning to RLHF.

Quantum computing concepts and their possible role in machine learning.

Reading papers, running experiments, and doing original research.

Near-term to speculative directions, labeled by confidence.

Building real products across industries and domains.

Roles, skills, and routes into AI work.

A durable method for understanding technologies that do not exist yet.