Your complete AI/ML learning path built to track learning, practice, and progress.
The full AI/ML landscape math, classical ML, deep learning, transformers, RL, agents, MLOps sequenced into 644 topics with open-access research-paper references. Track your progress in a beautiful, soothing interface that actually wants you to come back tomorrow.
the map
The whole field, on one canvas.
Every concept that matters from foundational math to autonomous agents — woven into a single connected map. Domains give it color; your progress gives it light. See what depends on what before you take a single step.
The full landscape
From linear algebra to RLHF, from BPE to FSDP every concept that actually matters in an AI/ML career, organized and sequenced.
Prerequisite graph
See what unlocks what. The knowledge graph shows the dependency structure of the entire field, painted with your progress.
Original papers attached
Each topic links to the seminal arXiv paper. Want to know where transformers came from? Two clicks to Vaswani et al.
the curriculum
Sourced from Stanford CS229/231n/224n, MIT OCW, fast.ai, NeetCode, William Fiset, GeeksforGeeks, Karpathy's Zero-to-Hero, Lilian Weng's blog, and over three hundred seventy arXiv papers every reference is free to read, no paywalls.
learn
Don’t memorize it feel it.
Topics carry visuals you can poke. Slide a parameter and watch the concept move intuition you build with your hands, not definitions you reread.
a model learning to fit · Visual
epoch 0 · MSE 0.758
Watch the curve bend to thread the points as the error falls that’s a model tuning its parameters. Slide the learning rate: too small and it crawls, too large and it overshoots and diverges. Several core concepts activations, attention, gradient descent carry a hands-on visual like this.
See it move
Key concepts come with interactive graphs slide a parameter and watch the idea change. Intuition you can poke, not just definitions.
Soothing by design
Dark, material, premium. Made for long study sessions generous spacing, low-contrast accents.
tensor code
Don’t just track it practice it.
Reading about softmax is one thing; writing a numerically stable one is another. Tensor Code drops you into an editor with more than ninety problems that matter from a one-line ReLU to a full transformer block and runs your code the instant you submit.
Training loss at three learning rates
The same model at three learning rates — a small one crawls down, a larger one races to the floor. The kind of multi-line graph Tensor Code draws straight from your solution.
Nothing to set up
Your code runs in a secure, isolated workspace no installs, no configuration. Open a problem and start typing.
Checked against real tests
Visible and hidden tests grade both correctness and the tricky edge cases a solution that looks right but breaks on a shape mismatch won't slip through.
Watch your own code learn
Many problems draw a live graph straight from your solution slide the learning rate, watch a decision boundary form, see the loss curve fall. The math stops being abstract.
Or follow a guided track:
Every problem is wired to a topic on the map so practice lands exactly where you’re learning, not in a vacuum.
track
Watch the effort add up.
Every topic ticked and every problem passed flows into your dashboard a map that brightens, a profile of where you’re strong and what’s next.
mastery by domain
activity
664 eventsTick a topic or pass a problem and the day lights up. A sample dashboard — yours fills in as you go.
Start the journey.
See the whole field. Start where you are. Make it stick.
Sign in to begin