Timescale renamed itself Tiger Data in June 2025. The database is still TimescaleDB, and the rename is part of a wider repositioning from time-series database to what the company now calls Agentic Postgres: TimescaleDB plus pgai, pgvectorscale, text search and a copy-on-write storage layer for forking databases. For learners the practical effect is that a decade of blog posts, Stack Overflow answers and tutorials use the Timescale name while the current material lives at tigerdata.com. There is no academy and no certification here, just documentation, tutorials and a set of Learn PostgreSQL guides, so I am reviewing them as reading material. Taken that way they are very good.
The Learn guides explain Postgres concepts such as partitioning, indexing, monitoring and downsampling in plain language, and they are worth reading even if you never install TimescaleDB. The tutorials use realistic datasets, from simulated IoT sensors to real-time financial data and the NYC taxi set, and they show hypertables, continuous aggregates and compression doing real work. The AI angle is pgvectorscale, an extension that adds a DiskANN-style index to Postgres and works alongside pgvector. It is released under the PostgreSQL licence, the docs are clear, and if you want to keep embeddings in the database you already run rather than adding a separate vector store, this is one of the better-documented routes. pgai, which calls models from inside the database, is newer, and I found its docs less settled. The licensing needs a clear explanation, because the docs do not give one up front.
TimescaleDB is split. The core is Apache 2.0, but the code in the tsl directory is under the Timescale License, which is source-available rather than OSI open source. That covers most of what makes the tutorials impressive: compression and the Hypercore engine, continuous aggregates, retention policies, advanced hyperfunctions and job scheduling. Self-hosting those features is free.
Offering TimescaleDB as a hosted database service is not allowed. For most teams that is fine, but it matters if you plan to build a platform, and the tutorials do not flag it. My other gripe is the lack of honest comparison. Nothing here tells you when native Postgres partitioning is enough, or when ClickHouse would beat TimescaleDB on pure analytics. That is a normal vendor blind spot, but for a company that teaches Postgres as well as this one does, it stands out. My take: excellent free reading on Postgres, time-series and vector search in Postgres.
Understand the licence before building anything hosted.