Everything from your first SELECT to writing CUDA kernels and shipping LLMs — SQL, Python, classic ML, deep learning, computer vision, NLP from scratch, GPU systems and MLOps. Rigorous, hands-on, and wired to real code.
From installing PostgreSQL to window functions, CTEs and query optimization — the data language every AI engineer must own.
Data types, control flow, functions, OOP and functional programming — the language the entire stack is written in.
NumPy, Pandas, Matplotlib, exploratory data analysis and association-rule mining, ending in a full Bike-Sharing project.
Big-O, core data structures, sorting, searching and the interview patterns that turn engineers into hires.
Pipelines, classification, regression, imbalance, tuning, SVMs, trees, ensembles, time series, unsupervised learning and MLflow.
The full Stewart sequence explained simply: functions & limits, derivatives and their rules, applications, integrals & techniques, infinite series, and multivariable & vector calculus — every idea backed by an interactive visual and worked exercises.
Aggarwal's Linear Algebra & Optimization for ML, from scratch: vectors & matrices as operators, linear transformations & systems, eigenvectors & diagonalization, SVD & matrix factorization, the linear algebra of similarity & graphs, and the optimization that trains every model — each idea shown geometrically and wired to real ML.
A full statistics course for machine learning: descriptive statistics, probability & Bayes, random variables & distributions, sampling & the Central Limit Theorem, estimation & maximum likelihood, confidence intervals, hypothesis testing & A/B tests, regression, Bayesian inference, and information theory — every concept with an interactive and worked exercises.
From tensors to training deep nets: initialization, activations, normalization, optimizers, scheduling, dropout and regularization.
CNNs → detection → segmentation → tracking → depth & 3D → optical flow, plus CLIP and Vision Transformers. The flagship track.
micrograd → backprop → n-grams → BPE → GPT-2 from scratch → fine-tuning → RAG → agents. Karpathy-grade, end to end.
MDPs, dynamic programming, TD learning, Q-learning & DQN, policy gradients, PPO, multi-armed bandits, and RLHF for LLM alignment.
The systems tab: tensor programming, tensor & model parallelism, CUDA and Triton kernels, and how LLM inference really works.
Local inference with vLLM, containerization with Docker, orchestration with Kubernetes, cloud on AWS SageMaker, MinIO data lakes, autoscaling, monitoring and drift detection.
The engineering employers hire for: REST APIs, databases at scale (replication, sharding, indexing), caching, message queues, load balancing & autoscaling, the CAP theorem, reliability patterns, AWS building blocks, the Designing Data-Intensive Applications core (storage engines, encoding, transactions & isolation, quorums, consensus, stream vs batch), and full system-design walkthroughs — how to build systems that serve millions of users.
The day-to-day craft every professional engineer is expected to have: the Linux command line, file permissions, SSH & remote access, Git version control, CI/CD with GitHub Actions, testing strategies, and logging & debugging — the workflow that turns code into shipped, reliable software.