Library

Coding

• Python for ML interviews100 min
• Calling LLM APIs in Python: async, streaming, retries100 min

• DSA patterns AI roles still ask150 min
• ML coding: implementing an LLM block in PyTorch180 min

• SQL and pandas for data rounds120 min
• Classic ML from scratch: logistic regression and k-means90 min

ML fundamentals

• Embeddings basics: word2vec, contrastive learning, what embeddings capture70 min

• Classification metrics60 min
• Probability and statistics for ML interviews95 min
• Gradient descent and backpropagation110 min
• Bias/variance, overfitting and regularisation90 min

• A neural network from scratch in NumPy75 min
• Diffusion models basics120 min
• Distributed training: DDP, FSDP/ZeRO, tensor and pipeline parallelism120 min

Transformers

• Self-attention from scratch120 min
• The decoder-only transformer (GPT forward pass)120 min

• Tokenization and BPE90 min
• Multi-head attention and positional encoding120 min
• Fine-tuning and LoRA95 min
• Attention variants: MQA, GQA, MLA and FlashAttention120 min
• Mixture of experts90 min
• RL for LLMs: PPO, GRPO and verifiable rewards180 min

• Pretraining objectives and scaling laws110 min
• Long context: RoPE scaling, sliding windows and lost in the middle75 min
• Multimodal and vision-language models120 min
• Alignment: instruction tuning, RLHF and DPO100 min
• Dataset engineering for fine-tuning105 min
• Distillation and synthetic data90 min

Inference

• Decoding: greedy, beam, temperature, top-k/top-p95 min

• KV cache90 min
• Quantization for LLM inference90 min
• Batching, continuous batching and LLM serving80 min
• Transformer math: parameters, FLOPs and memory estimates90 min
• Reasoning models and test-time compute120 min

• Speculative decoding and other inference accelerations90 min
• Why inference is memory-bound: GPU basics for LLM engineers150 min
• Small models and on-device inference120 min
• GPU programming basics: CUDA and Triton kernels150 min

RAG

• Retrieval for RAG: chunking, hybrid search, reranking120 min
• Embeddings and vector search100 min
• Evaluating RAG: retrieval metrics and faithfulness80 min

• Advanced RAG: query rewriting, reranking, agentic RAG, long context vs RAG100 min
• Rerankers and fine-tuning embedding models150 min

• Vector database internals and index selection110 min
• Document parsing and ingestion for RAG: PDFs, tables, OCR75 min

Agents

• Tool use and function calling65 min
• Agent loops: workflows vs agents120 min
• Structured outputs and validation75 min
• Agent reliability: evals, failure modes and observability95 min
• Prompt engineering as a discipline90 min

• Model Context Protocol (MCP)90 min
• Context engineering and agent memory80 min
• Multi-agent systems and orchestration90 min

• Voice and realtime agents90 min
• Computer-use and browser agents120 min

Evals

• Evals and error analysis150 min
• LLM-as-judge: building and validating judges75 min
• Eval datasets, benchmarks and evals in CI120 min
• Hallucinations: detection and mitigation90 min
• Design an eval: the live exercise120 min

• LLM observability and feedback loops60 min
• A/B testing and product sense for AI features90 min

• Statistics for LLM evals90 min

System design

• GenAI system design120 min
• Cost and latency engineering for LLM apps90 min
• RAG vs fine-tuning vs prompting: choosing the approach60 min
• Design an AI customer-support agent150 min
• Design enterprise document search / a knowledge assistant150 min

• Classic ML system design: recsys, search ranking, feature pipelines150 min
• Guardrails and safety: prompt injection, jailbreaks, output filtering80 min
• LLM and agent security90 min
• Shipping LLM features: prompt and model versioning, rollouts and migrations90 min
• PII, privacy and data handling in LLM apps105 min
• Design a ChatGPT-style assistant150 min
• Design a coding copilot150 min
• Model routing and cascades75 min
• Design an LLM inference and batching API150 min
• LLM gateways: rate limits, token budgets, retries and idempotency120 min
• AI product sense and scoping90 min

• LLMs in search and recommendations120 min
• Text-to-SQL and LLMs over structured data120 min
• Design an agent infrastructure platform150 min

AI coding round

• The AI-enabled coding round120 min
• Take-home projects and eval take-homes110 min

• Driving a coding agent in an interview75 min

Behavioral

• Behavioral interviews: STAR stories for AI engineers90 min
• The project deep-dive round60 min
• Story bank180 min
• Interview openers and closers90 min

• Company values rounds120 min
• Senior and staff behavioral signals120 min