ChromaDB Basics

Introduction

Chroma is an open-source, embedded vector database that runs in-process or as a server. Perfect for prototyping, local dev, and small/medium production loads.

Beginner analogy: SQLite for vectors — zero config, runs anywhere.

Understanding the topic

Core concepts to understand:

• Pip install or Docker run.
• Built-in embedding generation (or BYO).
• Persistent storage (DuckDB + Parquet).
• Python, JavaScript, REST APIs.

Syntax reference

Visual workflow / architecture:

import chromadb
client = chromadb.Client()
col = client.create_collection("docs")
col.add(documents=["..."], ids=["1"])
col.query(query_texts=["..."], n_results=3)

Real-world use

Heavy use in RAG tutorials, hackathons, prototypes. Many startups start on Chroma and migrate to Pinecone/Qdrant at scale.

Best practices

• Start with Chroma; migrate when you outgrow it.
• Use persistent client mode in production, not in-memory.

Hands-on exercise

Interview preparation — practice these questions:

• Q1. What is ChromaDB?
• Q2. When would you migrate off Chroma?
• Q3. Compare Chroma vs Pinecone.