← Back
Hugging Face Blog

Time-Series Models in the Streamhouse: A New Frontier and Use Cases

•12 min read•
#deployment#agents
Time-Series Models in the Streamhouse: A New Frontier and Use Cases
Level:Advanced
For:AI Engineers
✦TL;DR

IBM’s time‑series foundation models, which have amassed over 50 million downloads, are now available as native SQL functions on Confluent’s streaming platform, running on CPUs both in Confluent Cloud and on‑premises via Apache Flink. These models enable real‑time forecasting, anomaly detection, and semantic intelligence directly on streaming data without GPU requirements. Engineers can embed model outputs into downstream applications and AI agents, shortening the decision latency for use cases such as production line monitoring, retail stock‑out prevention, and payment fraud detection.

⚡ Key Takeaways

  • More than 50 million downloads of the IBM time‑series model family (four models) are now offered as built‑in functions in Confluent.
  • Models are deployed as native SQL functions within Apache Flink, allowing streaming inference without custom serving infrastructure.
  • Running on CPUs eliminates GPU dependency, reducing hardware cost and simplifying on‑premises deployment.
  • Integration is achieved via Confluent Cloud (fully managed) or Confluent Platform for Apache Flink on‑premises, using standard SQL calls.
  • Prerequisite: Access to a Confluent environment with Apache Flink enabled; models are not available outside this platform.
💡 Why It Matters

By bringing time‑series inference to the data‑in‑motion layer, engineers can deliver actionable forecasts and anomaly alerts at the moment events occur, dramatically improving the timeliness of automated decisions in production, retail, finance, and other real‑time domains.

✅ Practical Steps

  1. Use the provided SQL functions in Apache Flink jobs on Confluent to call the IBM time‑series models for forecasting or anomaly detection.
  2. Deploy the Flink job either in Confluent Cloud (no infrastructure to manage) or on‑premises via Confluent Platform to run the models on CPU resources.
  3. Incorporate the model outputs into downstream applications or AI agents by querying the enriched stream or Lightning Tables for low‑latency access.

Want the full story? Read the original article.

Read on Hugging Face Blog ↗

More like this

New in AWS Quick: More capable flows, an updated web experience, and agents that work together

AWS ML Blog•#agents

Three insights you may have missed from theCUBE’s coverage of the Dell Technologies New York Symposium

SiliconANGLE AI•#deployment

How Google Duplex evolved to also call businesses for you in Gemini

Google Research•#agents

Welcome Inkling 2 Mini, Cutting-edge Performance for your Local Setup

Hugging Face Blog•#llm

EXPLORE AI NEWS

Daily hand-picked stories on LLMs, RAG, agents and production AI — curated for engineers who ship.

BROWSE NEWS

GET THE WEEKLY DIGEST

Join engineers getting the Monday signal-over-noise AI breakdown. No spam, unsubscribe anytime.

LEARN AI ENGINEERING

Curated courses, research papers, repos and tutorials built for engineers leveling up in AI.

START LEARNING