Seeknal

Seeknal provides a command‑line interface that unifies data pipelines, feature stores, and natural‑language querying for AI/ML workflows. Users describe transformations in YAML, Python decorators, or a mix of both, and the tool compiles them into a single execution graph that can materialize results to PostgreSQL or Iceberg. A built‑in four‑stage workflow—draft, dry‑run, apply, and query—ensures changes are previewed before they affect production, reducing silent failures.

The platform also includes a feature‑store component that supports point‑in‑time joins, watermark‑based incrementals, and automatic versioning, allowing both batch and real‑time serving with optional TTL. Dashboards can be published, and an integrated agent can answer natural‑language questions by referencing the schema and pipeline metadata, turning insights into actions such as triggering jobs, sending alerts, or exposing APIs.

Seeknal is packaged as a single Python library, supporting DuckDB, PostgreSQL, and Iceberg engines, and works with AI providers like Gemini, Ollama, and OpenAI. It aims to simplify the stack for data and AI/ML engineering by consolidating pipeline authoring, feature management, and query handling into one cohesive CLI.