NeMo Curator

NVIDIA NeMo Data Designer is an open-source framework for generating synthetic datasets from scratch or from seed data. It builds datasets column by column using sampler columns, LLM-generated text or structured columns, image and other multimodal models, existing seed datasets, and dependency-aware field generation. Workflows can connect to local or remote MCP servers to use external tools and capture interaction traces. Outputs can be previewed, resumed, and monitored, then validated or scored with Python, SQL, custom validators, and LLM judges. The framework also supports plugins for custom columns, seed readers, and processors, and includes a command-line interface for configuring model providers and models. It is distributed as the `data-designer` Python package and its source repository is licensed under Apache License 2.0. The library includes optional aggregate telemetry for model names and token counts, which can be disabled with `NEMO_TELEMETRY_ENABLED=false`.

NeMo Anonymizer is a data-anonymization tool for sensitive information, including agent logs and accidentally leaked credentials.

NeMo Gym is an AI reinforcement-learning tool for defining reward functions as an alternative to supervised fine-tuning on generated golden trajectories. It is used to train models toward a target agent behavior by evaluating their generated trajectories rather than relying only on prebuilt example traces.

DatologyAI is a data curation service for AI model builders. It processes proprietary enterprise data, public web data, and licensed data for pre-training and mid-training on open models or privately deployed foundation models. Its four-stage pipeline cleans poorly formatted, empty, short, or evaluation-contaminated data; curates it by quality, task, and taxonomy; creates synthetic data for task relevance and diversity; and composes sources into staged training datasets. The company also describes seeded rephrasing methods for generating synthetic data at trillion-token scale.

NAMS is a hosted, graph-native memory service from Neo4j Labs for LLM agents. Backed by Neo4j Aura, it captures messages, entities, tool calls, and reasoning traces into workspace-isolated graphs, with short-term conversation memory, long-term knowledge, synthesized observations, entity merging, hybrid vector and text search, and REST and MCP interfaces. Agents connect with a workspace API key, after which NAMS extracts and embeds memory automatically and supports graph, semantic, and three-tier context retrieval. Its skill-distillation feature runs a seven-stage Scope → Extract → Consolidate → Synthesize → Gate → Package → Publish pipeline that turns a scoped portion of the memory graph into a signed, provenance-grounded SKILL.md package. The process includes grounding, coverage, specification linting, PII redaction, human sign-off, typed procedure graphs, cryptographic attestation, composition of shared sub-procedures, and drift detection for stale or contradicted claims. A dashboard provides workspace and skill-distillation interfaces; the service also offers plugins for Claude Code, Gemini CLI, Codex, and OpenCode, and supports bring-your-own LLM keys or an external Neo4j deployment.