Documentation
Guides, API references, and examples for the Portiere SDK and Cloud platform.
Quick Install
New to Portiere? Follow the Getting Started guide to run your first mapping pipeline in 5 minutes.
Getting Started
Getting Started
Install the SDK and run your first mapping pipeline in 5 minutes.
API Reference
Complete reference for portiere.init(), Project class, and all public methods.
Configuration
LLM providers, thresholds, knowledge backends, engines, and YAML configuration.
Operating Modes
Local execution mode with intent-based configuration.
Database Connections
Connect to PostgreSQL, MySQL, SQLite, and other databases as data sources.
Advanced Guides
Building the Knowledge Layer
Programmatic usage of build_knowledge_layer() for each vector store backend with step-by-step examples.
Knowledge Layer
9 search backends: BM25s, FAISS, Elasticsearch, ChromaDB, PGVector, MongoDB, Qdrant, Milvus, and Hybrid with RRF fusion.
LLM Integration
BYO-LLM guide: OpenAI, Anthropic, Azure, Bedrock, and Ollama.
Data Models
SchemaMapping, ConceptMapping, approval workflows, and export formats.
Custom Standards
Define your own clinical data standard as a YAML file and use it with the full mapping pipeline.
Pipeline Architecture
The 5-stage Ingest, Profile, Schema Map, Concept Map, ETL + Validate pipeline.
Error Handling
Structured exceptions, error recovery patterns, and retry strategies.
Vocabulary Setup
Download, prepare, and index OMOP vocabularies (SNOMED, LOINC, RxNorm, ICD10CM) for local use.
Elasticsearch Backend
Use Elasticsearch as a full-text search backend for concept matching with existing ES infrastructure.
Hybrid Search
Combine multiple search backends (BM25s + vector stores) with RRF fusion for maximum accuracy.
Mapping Review Workflow
Approve, reject, override, export to CSV, and reload reviewed mappings in the SDK.
Roadmap
Notebook Examples
Explore 9 hands-on Jupyter notebooks covering CSV, Parquet, database sources, concept mapping, ETL generation, local mode, cloud collaboration, and end-to-end FHIR R4 pipelines.
View Notebooks on GitHub →