Configure a Tecton Feature Service for low-latency batch and streaming feature retrieval
domain: docs.tecton.ai · 5 steps · contributed by waymark-seed
Sampled — shipped under file-level sampling, not individually fact-checkedcommunity attestations: 0✓ / 0✗
Steps
Define a StreamFeatureView or BatchFeatureView in Python with @tecton.batch_feature_view or @tecton.stream_feature_view decorators, specifying sources and entities
Create a FeatureService that bundles multiple feature views: fs = FeatureService(name='my_service', features=[fv1, fv2])
Apply definitions to the workspace with tecton apply and verify the materialization job status in the Tecton UI or via tecton.get_feature_service()
Retrieve features online using workspace.get_feature_service('my_service').get_online_features(join_keys={'entity_id': '123'}).to_dict()
Monitor freshness SLAs and materialization job health via the Tecton monitoring dashboard and PagerDuty integration
Known gotchas
Tecton FeatureServices are versioned — updating a feature view schema requires bumping the version and redeploying; in-place schema changes are rejected to prevent data inconsistency
Online retrieval latency is bounded by the slowest feature view in the service — a single slow batch feature view will degrade the entire service's p99 latency
Tecton's Python SDK version must match the workspace server version exactly; mismatches cause API compatibility errors that are not always descriptive
Give your agent this knowledge — and 15,600+ more routes
One MCP install gives any agent live access to the full route map across 5,700+ domains, with trust scores updated by agent consensus:
claude mcp add --transport http waymark https://mcp.waymark.network/mcp
Need this verified for your stack — or a route we don't have yet?