Fennel provides a platform for authoring, computing, and managing both real-time and batch machine learning pipelines using a CDC-aware engine that enables automatic incremental computation. This technology allows businesses to efficiently handle data changes, ensuring data freshness while minimizing compute costs and eliminating the need for extensive data engineering expertise.
Funding
$6M raised to dateRaised to date based on public sources. This may differ from the amount the company actually raised and is based only on what is publicly available on the internet.
Founders
Product
Problem
Building and managing machine learning pipelines for real-time and batch processing requires significant data engineering effort and expertise. Traditional approaches often involve recomputing entire pipelines even when only a small portion of the underlying data has changed, leading to high compute costs and stale data.
Solution
Fennel provides a platform for authoring, computing, storing, serving, monitoring, and governing both real-time and batch machine learning pipelines. Its core innovation is a Change Data Capture (CDC)-aware engine, written in Rust, that enables automatic incremental computation. This approach ensures data freshness while significantly reducing compute costs by only processing data proportional to the changes. Fennel eliminates the need for specialized data engineering skills by offering a Python-based interface and fully-managed infrastructure, allowing data scientists and ML engineers to focus on model development and deployment.
Target Audience
Fennel is designed for machine learning and data teams that need to build and manage real-time and batch ML pipelines efficiently, with a focus on data freshness, cost optimization, and ease of use.
Features
- Incremental computation engine that automatically processes only changed data, reducing compute costs.
- Python-based interface with Pandas integration, eliminating the need for domain-specific languages (DSLs) and simplifying pipeline authoring.
- Automatic backfilling of pipelines upon declaration, removing the need for manual scripting.
- Fully-managed infrastructure that handles scaling, reliability, and security, including SOC2 compliance and deployment within a customer's VPC.
- Feature repository for creating, sharing, and reusing standardized features across multiple use cases.
- Strong typing, immutability, and versioning to prevent data quality issues and offline/online skew.
- Built-in data quality tooling, including unit testing, compile-time validation, and data expectation monitoring.
- REST API for querying features.
- Built-in connectors to ingest data from various sources.