LLM Frameworks & Orchestration
- LangChain / LangGraph: The standard for chaining LLM calls, tools, and agents with graph-based workflows.
- LlamaIndex: Specialized for data ingestion, indexing, and RAG pipelines.
- Haystack: End-to-end NLP framework with strong document search capabilities.
Vector Databases
- Pinecone: Managed, serverless vector search at scale.
- Weaviate: Open-source with hybrid search and GraphQL API.
- Chroma: Lightweight, developer-friendly for prototyping.
- pgvector: PostgreSQL extension — great when you already use Postgres.
Development & Deployment
- Cursor / GitHub Copilot: AI-assisted coding in the IDE.
- v0 / Bolt: AI-powered UI generation from prompts.
- Modal / Replicate: Serverless GPU inference for custom models.
- Weights & Biases: Experiment tracking for ML and fine-tuning.
Evaluation & Safety
Tools like LangSmith, Braintrust, and Guardrails AI help you evaluate LLM outputs, trace agent decisions, and enforce safety policies before responses reach users.