ML feature store template

How a feature store serves the same features to training and real-time inference: batch and streaming pipelines, offline and online stores.

Use this templateFree, opens in your browser. No sign-up.

ML feature store template
Feature store (e.g. Feast, Tecton)
Why a feature store
Postgres: orders, users
Kafka: click + payment events
S3 data lake (Parquet)
Batch transforms: Spark / dbt, nightly
Streaming transforms: Flink, seconds
Registry: definitions, owners, versions
Offline store: warehouse, full history
Online store: Redis, latest values
Data scientists: discover + reuse features
Training: point-in-time correct joins
Fraud scoring API: < 10 ms lookups
Example feature definitions
Feature
Entity
Source
Freshness
TTL
orders_last_7d
user_id
Postgres (batch)
Daily
7 days
avg_basket_value_30d
user_id
Data lake (batch)
Daily
30 days
failed_payments_1h
card_id
Kafka (stream)
< 5 s
1 hour
new_device_flag
session_id
Kafka (stream)
< 5 s
24 hours
merchant_chargeback_rate_90d
merchant_id
Postgres (batch)
Daily
90 days
Same feature logic for training and serving
No leakage: point-in-time joins
Teams reuse features instead of rebuilding
ML feature store
materialize
log
browse
deploy model
The ML feature store template as it opens in Kelvo: 11 boxes, 3 sticky notes, 1 table and 11 links, in 2 sections.

What's on the board

Feature store (e.g. Feast, Tecton)
Registry: definitions, owners, versions · Offline store: warehouse, full history · Online store: Redis, latest values
Why a feature store
Same feature logic for training and serving · No leakage: point-in-time joins · Teams reuse features instead of rebuilding

Table columns: Feature, Entity, Source, Freshness, TTL.

Also on the board

  • Data scientists: discover + reuse features
  • Postgres: orders, users
  • Batch transforms: Spark / dbt, nightly
  • Training: point-in-time correct joins
  • Kafka: click + payment events
  • Streaming transforms: Flink, seconds
  • Fraud scoring API: < 10 ms lookups
  • S3 data lake (Parquet)
  • Example feature definitions

How to use this template

  1. Click Use this template. It opens in the Kelvo app as a new board, saved in your browser.
  2. Double-click any box to change its text. Replace the examples with your own content.
  3. Sketch, then convert. Draw boxes and arrows with the pen, select the strokes and click Convert sketch: they become real boxes and links.
  4. Ask Kelvo right on the board (Pro). With Pro, type @kelvo and what you want in any box or text, for example “@kelvo add error handling”, and press Enter. The change is made there and the request disappears; ⌘Z undoes it.
  5. Keep the prompt next to the diagram. Add a code block from the Insert (+) menu for prompts, schemas or config, right beside the boxes they belong to.

Good for

  • ai
  • machine learning
  • feature store
  • mlops
  • data engineering
  • Explaining an LLM feature's architecture to engineers and stakeholders
  • Deciding between prompting, retrieval and fine-tuning
  • Planning evaluations and risks before launch

Related templates

  • MLOps lifecycle template previewMLOps lifecycle

    The machine learning loop from problem framing to monitored production model, with retraining when data drifts.

  • Data labeling pipeline template previewData labeling pipeline

    Who does what when labeling training data: guidelines, model pre-labels, double annotation, QA on agreement and retraining.

  • ML experiment tracker template previewML experiment tracker

    Log every model experiment with its hypothesis, change, metrics and decision, so the team knows what worked and what to try next.

  • RAG pipeline template previewRAG pipeline

    Retrieval-augmented generation end to end: ingest and embed documents offline, then retrieve, rerank and answer with citations.

  • LLM agent loop template previewLLM agent loop

    How a tool-using agent plans, calls tools, observes results and updates memory until the goal is met, with guardrails.

  • Data lakehouse architecture template previewData lakehouse architecture

    Sources, ingestion, bronze/silver/gold tables on object storage, the consumers that read them, and the governance around it all.

All ai and machine learning diagram templates · All templates

Use the ML feature store template