Intelligence data,
end to end.
Hermes is a foundational intelligence data platform. Acquire, validate, normalize, store and serve datasets — with provenance baked into every row.
pip install hermes-pltPython 3.11+ · pandas, polars, arrow & duckdb friendly
One pipeline. Every source.
From raw fetch to production-ready features in a single, repeatable pipeline.
Acquire
Pull data from APIs, CSVs, JSON and public datasets through a unified connector layer with retries, rate limiting and a shared cache built in.
Validate
Check integrity, schema and provenance before anything dirty ever reaches a dataset. Catch bad data at the source, not downstream.
Normalize
Bring messy sources into one canonical shape with consistent codes, units and timestamps across every connector.
Serve
Query, export and feed features into models — lineage tracked every step of the way, ready for ML and analytics.
Provenance you can point at.
Every Hermes dataset carries metadata, provenance and lineage. Know where data came from, how it changed and what depends on it — without a separate system.
- Unified connector contract for every source
- Tiered, dependency-aware feature resolution
- Parquet-backed caching with per-source TTLs
- Works with pandas, polars, arrow & duckdb
from hermes import Hermes
hermes = Hermes()
# Fetch global macro + country risk
world = hermes.world_bank.fetch(
indicator="NY.GDP.MKTP.KD.ZG"
)
# Compute economic features with lineage
features = hermes.economic_features.compute(
dataset=world
)
stats = hermes.cache.stats()
print(f"cached: {stats['hits']} hits")Ship trustable data today.
Get the full pipeline running in under a minute. No accounts, no API keys required to start.