The 14-day plan, day by day
The fourteen days, and what you can answer after each
The core topics over twelve days, then two days of interview practice. AI System Design, the AI Engineer Stack and the three cloud sessions are yours to keep.
Every question has a written model answer, and the ones being asked this year are in there.
1How a model turns your text into tokens and predicts the next oneQ&A
2Attention, Mixture of Experts, the KV cache, why it hallucinatesQ&A
3Context engineering: what to put in the window and what to leave outQ&A
4Embeddings, similarity and the ANN indexes behind vector searchQ&A
5Chunking and hybrid search — deciding what can ever be retrievedQ&A
6Query rewriting, reranking, and measuring retrieval on its ownQ&A
7Tools, function calling, structured outputs and MCPQ&A
8Agents: the ReAct loop, memory, checkpointing, stop conditionsQ&A
9Multi-agent: handoffs, supervision, and when one model is betterQ&A
10Evaluation and guardrails — Recall@k, LLM-as-a-judge, red teamingQ&A
11Serving: continuous batching, speculative decoding, model routingQ&A
12Running it: tracing, circuit breakers, drift, cost per requestQ&A
13Mock interviews. You answer out loud; the coach scores deliverytools
14Resume scored and rebuilt, roles matched, and a full reviewtools
