Embedded analytics · running live

Live demos

These demos are small, production-style data products running in public. Each uses a live data source, processes and stores the incoming data, models it for analysis, and serves the result through an embedded application experience.

Live editing activity across English, Portuguese, and German Wikipedia.

Connecting to the live data stream…

Freshness is measured from the event’s source timestamp rather than the time this page was opened.

Embedded within Northwind, a demonstration host application

Northwind Home Projects Insights

Under the hood

See how the pipeline ingests, stores, queries, and serves live data, including end-to-end latency, operational health, and the API behind the embedded experience.

  1. 01
    Ingest

    A lightweight ingestion service consumes Wikimedia’s public event stream, automatically reconnects following interruptions, and filters test events before processing.

  2. 02
    Store

    Raw events and per-minute aggregates are written to an analytical database, with validation for duplicate records and missing fields.

  3. 03
    Serve

    An API exposes the metrics required by the embedded widget, keeping the presentation layer isolated from the upstream source.

One event’s journey, measured
Wikipedia event created
0 s
Persisted by the pipeline
n/a
Delivered to the application
n/a

Stored latency reports the median and 95th percentile over the previous minute, measured from each event’s source timestamp. Delivery latency measures the age of the latest event when the application last retrieved it.

Architecture