Loading
Pick a focused question that fits your time, stack, and interview goal.
20 results across 1 active filter
Page 1 of 1
Showing the AI Engineering curriculum. Technology tags describe each drill's actual implementation context; shared drills can belong to more than one track.
Diagnoses an AI regression by decomposing request stages, configuration changes, token growth, retries, routing, tools, retrieval, caching, and provider behavior.
Compares managed model APIs with operating open-weight models across control, expertise, privacy, scale, latency, cost, and portability.
Separates runtime model use from model training, fine-tuning, prompt design, and retrieval-augmented generation as different engineering levers.
Explains how decoding settings reshape token selection and why they must be tuned against task-specific evaluation rather than folklore.
Balances AI quality, tail latency, and cost through explicit product thresholds, model choice, context control, caching, routing, and measurement.
Handles token-budget overflow and incomplete generation through explicit prioritization, rejection, retrieval, summarization, continuation, and validation.
Separates model and provider outcomes, applies bounded recovery, validates outputs, and exposes useful status and telemetry.
Chooses a model from task-specific evaluation, operating constraints, safety, latency, cost, context, modality, and provider requirements.
Designs a bounded and reversible first AI feature with measurable value, controlled data, validation, evaluation, observability, fallback, and human oversight.
Treats model changes as controlled releases using versioned configuration, fixed evaluations, shadow or canary traffic, monitoring, and rollback.
Explains tokens as model-specific text units and connects tokenization to context limits, output limits, latency, cost, and text behavior.
Defines hallucination as unsupported or incorrect model output and connects it to evidence, task design, evaluation, and user-visible uncertainty.
Describes the context window as the bounded request state containing instructions, conversation, retrieved evidence, tool results, and output capacity.
Defines a large language model through next-token prediction and connects that mechanism to useful language behavior and its limits.
Defines embeddings as model-produced vectors used for similarity and contrasts representation with text generation.
Separates what a model can perform from whether the complete AI feature behaves dependably under real inputs, failures, and constraints.
Identifies tasks better served by deterministic code, search, rules, or conventional models because of correctness, latency, cost, or control.
Uses task bounds and evaluation evidence to choose smaller models for lower latency, cost, capacity, privacy, or local execution.
Connects probabilistic decoding and changing service conditions to output variation, reproducibility limits, and appropriate testing.
Places validation, authorization, limits, state transitions, and approval outside the model while acknowledging that guardrails reduce rather than eliminate risk.