AI & ML Tech Glossary

AI red teaming is adversarial testing of AI systems, especially LLM apps, to uncover safety and security failures like jailbreaks, data leakage, harmful outputs, and...

Evals are systematic tests for LLMs and LLM apps that score quality, safety, and reliability using datasets, rubrics, automated checks, and sometimes human review. They...

LLM observability instruments and monitors LLM applications with traces, logs, and metrics that capture prompts, retrieval context, tool calls, tokens, latency, cost, and quality signals....

Model parallelism splits a single neural network across multiple GPUs or accelerators so each device computes part of the model. It enables training and serving...

An agentic workflow is a repeatable sequence of planning, tool calls, evaluation, and iteration that an AI agent follows to complete a goal. It makes...

Self-consistency decoding samples multiple reasoning outputs from an LLM and selects the most consistent final answer using voting or aggregation. It often improves accuracy on...

An agent swarm is a multi-agent system where several AI agents collaborate in parallel—using role specialization, coordination, and cross-checking—to plan and execute complex tasks more...

LLM guardrails are runtime controls—input/output checks, tool permissions, policy enforcement, and monitoring—that constrain a language model to be safer, more reliable, and compliant in real...