# airbnb/aerosolve

A machine learning package built for humans.

Repository: https://github.com/airbnb/aerosolve
Canonical: https://ross.abutalabs.com/products/aerosolve
Homepage: http://airbnb.github.io/aerosolve/
Language: Scala
License: Apache-2.0
License Family: permissive
Last push: 2025-11-06T00:53:40+00:00

## Health v2 (maintenance only)
Score: 45/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 50, release rhythm 8, longevity 100
- inputs: {"age_days": 4131, "days_push": 301, "days_rel": 415, "gap_med": 293, "n_releases_24m": 2}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 4808, forks 563 (observed 2026-08-28T04:09:00.257011+00:00)

## What it is
Aerosolve is a machine learning library from Airbnb built for human-friendly, interpretable modeling on the JVM. It provides a thrift-based feature representation, a feature transform language, debuggable additive models, Scala training code, and a lightweight Java inference API.

## Use cases
- build interpretable machine learning models with sparse features
- train ranking models for search results
- model pricing and demand prediction
- debug and visualize feature weights
- engineer geo-based and feature interaction transforms
- serve lightweight Java inference in production
- rank or order images by content

## When to choose
- you need interpretable, debuggable models over sparse human-readable features
- you work on the JVM and want Scala training with lightweight Java inference
- your problem involves search ranking, pricing, or demand modeling
- you want fine-grained control over feature transforms and quantization

## When to avoid
- you need deep learning on dense inputs like raw pixels or audio
- you want a actively developed library with modern ecosystem support
- you need GPU acceleration or Python-based tooling
- your team is not comfortable with Scala/Java

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, nlp, image-processing
- domain: machine-learning, data-science, developer-tools
- platform: jvm
- tags: interpretable-ml, feature-engineering, ranking, sparse-features, additive-models, scala, java

## Member repositories
- airbnb/aerosolve (main) score 45

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:09:00.257011+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T18:18:36.175349+00:00, confidence not recorded.
  - readme: https://github.com/airbnb/aerosolve (fetched 2026-08-28T04:09:00.257011+00:00, sha c461a5ca4f15)
  - homepage: http://airbnb.github.io/aerosolve/ (fetched 2026-08-29T09:01:35.390779+00:00, sha c54da9ea55d7)
- Data as of 2026-08-30T08:39:29.467469+00:00.