Read short lessons, run live calculators on your own numbers, and test yourself with an adaptive learner model that knows what you're weak at — then drills it.
Type numbers, see the result update instantly.
From programming and probability to transformers, RAG and AI agents — every idea is connected to its formula, its calculator and a quiz that remembers how you did.
Students, developers moving into AI, and data scientists who want a clear map of what to learn.
Every topic has prerequisites, so you always know what to study next — and why.
Item Response Theory estimates your ability per subject with confidence intervals.
SM-2 schedules reviews so you retain what you learn instead of forgetting it.
From first principles to production inference. Pick a starting point or follow the path.
Python, SQL, algorithms, Git, REST, Docker, CI/CD.
🏗️Pipelines, ETL/ELT, Kafka, Spark, Airflow, warehouses.
🗄️PostgreSQL, MongoDB, Redis, vector and graph databases.
📐Vectors, eigenvalues, decompositions, SVD, PCA.
∫Derivatives, gradients, chain rule, optimization.
🎲Random variables, distributions, Bayes, conditioning.
📊Hypothesis tests, confidence intervals, regression.
📡Entropy, cross-entropy, KL divergence, mutual info.
🌲Trees, boosting, SVM, clustering, reinforcement learning.
🧠Networks, optimizers, regularization, CNNs, RNNs, GANs.
✨LLMs, transformers, fine-tuning, LoRA, quantization.
📚Chunking, embeddings, hybrid search, reranking, Graph RAG.
🤖Tool calling, memory, multi-agent systems, MCP, LangGraph.
👁️Detection, segmentation, CLIP, diffusion models.
🗣️Tokenization, NER, summarization, translation, embeddings.
🚀GPUs, batching, vLLM, TensorRT, ONNX, Triton.
🔐Prompt injection, poisoning, adversarial attacks, secure RAG.
⚖️Fairness, bias, transparency, oversight, governance.
🔍SHAP, LIME, partial dependence, counterfactuals.
📊ML metrics, LLM groundedness, hallucination, agent evals.
📈ARIMA, Prophet, LSTM, transformers, anomaly detection.
Not just reading. Every concept has a calculator, a quiz, and feedback that adapts to you.
Eigenvalues, Bayes, KV-cache memory, LoRA size — enter your own values and see results instantly.
Missed questions come back sooner; mastered ones move further out. Feedback with explanations.
Item Response Theory estimates ability θ per subject with confidence intervals and calibration.
Explain a concept in your own words. Get graded by keyword coverage or by your own LLM.
A live knowledge graph of all 66 topics, colored by mastery, with the frontier highlighted.
Python (via WebAssembly) and SQL (via SQLite). Runs entirely in your browser — nothing uploaded.
The exact statistical models and their limitations are documented on a public page.
Add your own topics in-browser, preview live, export as JS. Everything stays local.
Seven stages that build on each other. Scroll or drag to explore.
Python, SQL, Git, Docker, CI/CD.
→Pipelines, Kafka, Spark, warehouses, vector DBs.
→Linear algebra, calculus, probability, statistics.
→Regression, trees, SVM, clustering, RL.
→Networks, optimization, CNNs, RNNs, transformers.
→Training, fine-tuning, retrieval, evaluation.
→Inference, serving, security, responsible AI.
Three steps per topic, repeated across the roadmap.
A short explanation with the key formulas on clear cards. No walls of text.
Change the numbers in the calculator and see the answer update instantly.
Answer a question, log your confidence, get feedback, and progress builds automatically.
No sign-up. No credit card. Just open the platform and start with any subject.
Open the learning platform