All Tracks

7 structured tracks covering every ML interview topic. Track 7 runs in parallel with your current focus track.

Probability, linear algebra, information theory, classical ML, and evaluation — the bedrock of every interview.

From perceptrons to state-space models — neural network architectures old and new.

Tokenization through evaluation — the full NLP/LLM stack from bigrams to GPT.

Loss functions, optimizers, schedules, mixed precision, distributed training, and finetuning.

KV-cache, attention kernels, quantization, serving, speculative decoding, and production ML.

Multimodal models, vision transformers, self-supervised learning, retrieval, safety, and paper reading.

Implement from scratch, NumPy/PyTorch exercises, and ML algorithm problems — runs in parallel with all other tracks.