Frontier Atlas

The math of vectors and transformations — the language of ML, graphics, and quantum.

Continuous change and the machinery of optimization.

Reasoning under uncertainty — the backbone of all inference.

Logic, sets, combinatorics — the math of countable structures.

Nodes and edges — networks, routing, dependencies, everything connected.

What information is, how much fits, and the limits of compression.

The mathematics of computation and proof — where logic meets programs.

The math of structure and composition — abstraction's abstraction.

Computing real-valued answers when exact ones are impossible.

How to compute efficiently — the craft at the core of all software.

The abstract machines that define what computation even is.

What can and cannot be computed — the hard edges of the possible.

How hard problems really are — P vs NP and the cost of solving.

Error-correcting codes — making data survive noisy channels.

Mathematically proving software correct instead of just testing it.

The software that manages hardware and runs everything else — the layer between bare silicon and every program you use.

Turning human-readable code into machine instructions.

Designing the languages we think and build in.

Storing, indexing, and querying data at scale and speed.

Making many machines act as one, despite failure.

Moving data across the world reliably — the plumbing of the internet.

Filesystems, LSM trees, B-trees — how bits persist.

Running many isolated machines on one — the cloud's foundation.

Packaging and scheduling workloads — Docker, Kubernetes, and beyond.

Computing with hard deadlines — where late means wrong.

Squeezing maximum compute from many cores — CUDA, MPI, OpenMP.

Portable, near-native execution everywhere — the universal runtime.

Gates and signals — the bedrock beneath every chip.

How CPUs are organized — pipelines, caches, memory hierarchies.

Designing processors from RTL to tapeout.

Massively parallel silicon — the engine of graphics and AI.

Reconfigurable hardware — Verilog, VHDL, computing you rewire.

Custom chips for one job, done supremely well.

Software living inside devices, close to the metal.

Designing the physical boards that hold the silicon.

DRAM, NVM, storage-class memory — where speed meets persistence.

TPUs, NPUs — silicon built specifically to run neural networks.

Managing and converting electrical power efficiently.

Systems that improve at a task by learning patterns from data, instead of being explicitly programmed.

Many-layered neural networks — the engine of the modern AI wave.

Teaching machines to see and understand images and video.

Machines that read, write, and reason in human language.

Learning by trial, reward, and consequence.

Many AIs coordinating, competing, and negotiating.

Deep learning over networks and relational structure.

Diffusion, GANs, VAEs — machines that create.

Going beyond correlation to actual cause and effect.

Modeling uncertainty explicitly — priors, posteriors, beliefs.

Training models across devices without centralizing data.

Learning to learn — adapting fast from little data.

Models that fuse text, image, audio, and more.

Program synthesis and AI that writes and reasons about software.

Predicting what you'll want next — the engine of modern platforms.

Modeling and predicting sequences through time.

Making powerful AI systems do what we actually want.

Understanding what's happening inside the black box.

Fusing neural learning with symbolic reasoning.

Running models on microcontrollers and tiny devices.

The engineering of training, shipping, and serving models reliably.

Giant general-purpose models and the science of scaling them.

Machines that sense, decide, and act in the physical world — the union of perception, control, and mechanism.

Computing how a robot should move — safely and optimally.

How a robot maps the world and finds itself in it.

Fusing vision, lidar, and radar into understanding.

The unsolved art of robotic hands picking things up.

Robots made of compliant, bio-inspired materials.

Many simple robots producing complex collective behavior.

Robots that work safely and naturally alongside people.

Precision machines inside the operating room.

Self-driving cars, trucks, and the full self-driving stack.

Flying robots — control, autonomy, and swarms in the air.

Robots that walk, run, and balance on legs.

Robots that learn skills from demonstration and practice.

Rovers, arms, and autonomous craft beyond Earth.

The math and engineering of keeping secrets and proving identity.

Proving something is true while revealing nothing else.

Computing directly on encrypted data without decrypting it.

Many parties computing a result without revealing their inputs.

Hardware enclaves that protect code even from the OS.

Breaking systems legally to find holes before attackers do.

Taking software apart to understand what it really does.

Reconstructing what happened from digital evidence.

Turning memory bugs into control of a machine.

Extracting secrets from power, timing, and electromagnetic leaks.

Encryption that survives quantum computers.

Synthesizing images — ray tracing, rasterization, shaders.

Building interactive worlds and the engines that run them.

Simulating physical reality fast enough to interact with.

Digitally shaping and analyzing sound.

Generating and analyzing music with code and AI.

Software that redefines what a camera can capture.

Generating and efficiently encoding moving images.

Building immersive and augmented experiences.

Computing you can feel — touch as an output channel.

Algorithms that create endless content from rules.

Studying and structuring how humans use interfaces.

Making computing usable by everyone, regardless of ability.

Reading the body as an input device.

Using gaze as a signal and a control.

Direct channels between brains and machines.

Systems that sense and respond to human emotion.

Talking to machines and having them understand.

Reading the code of life with computation — turning biological data into biological understanding.

Modeling how brains compute.

Decoding genes and the proteins they build.

Designing medicines in silico before the lab.

Predicting 3D shape from sequence — the AlphaFold revolution.

Simulating molecules and reactions from first principles.

Discovering new materials with data and simulation.

Simulating the planet's systems to see the future.

Simulating how fluids and gases flow.

Simulating stress, heat, and deformation in structures.

Modeling stars, galaxies, and the evolution of the cosmos.

Simulating fusion, fission, and the fourth state of matter.

Solving physics that pen and paper can't.

Computing for medicine — imaging, records, diagnostics, devices.

Algorithms for trading, risk, and the movement of money.

Encoding and automating the legal system.

Software that teaches and adapts to each learner.

Precision farming and computational agriculture.

Computing aimed at decarbonizing the world.

Mission software, telemetry, and the new space economy.

Remote sensing, satellite imagery, and digital twins of Earth.

Smart grids, batteries, and forecasting the flow of power.

Optimizing how things move across the world.

Computing with quantum mechanics — using superposition and entanglement as the raw material of computation.

Unhackable communication via quantum states.

Chips that compute like brains — spiking, event-driven, low-power.

Computing and storing data in molecules.

Computing with light instead of electrons.

Emergent intelligence from many simple agents.

Programming living cells like machines.

Living virtual replicas of physical systems and cities.

The consensus engines under decentralized networks.

Code that runs trustlessly on a blockchain.

Scaling and privacy for blockchains via zero-knowledge.

Rebuilding financial primitives without intermediaries.

Storing data across a network with no central owner.

Designing incentives that secure decentralized systems.

AlphaFold, drug design, genomics — AI decoding life.

Physics-informed neural networks and learned simulation.

ML compilers and inference engines — making models run fast.

Adversarial ML and AI for threat detection.

Co-designing silicon and models — TPUs, neuromorphic chips.

Privacy-preserving and federated machine learning.

Machine learning on quantum hardware — early but tantalizing.

NeRF, Gaussian splatting, generative 3D worlds.

Adaptive, intent-predicting interfaces.

Surgical robots, prosthetics, and exoskeletons.