# flyteorg/flyte

Dynamic, resilient AI orchestration. Coordinate data, models, and compute as you build AI workflows.

Repository: https://github.com/flyteorg/flyte
Canonical: https://ross.abutalabs.com/products/flyte
Homepage: https://flyte.org
Language: Go
License: Apache-2.0
License Family: permissive
Topics: flyte, machine-learning, golang, scale, workflow, data-science, data-analysis, kubernetes, orchestration-engine, mlops, grpc, python, production, declarative, fine-tuning, llm, hacktoberfest, agentic, ai-agents, ai-development-tools
Last push: 2026-08-26T21:54:51+00:00

## Health v2 (maintenance only)
Score: 95/100 (v2, computed 2026-09-03T02:39:23.370411+00:00)
- activity 99, release rhythm 87, longevity 100
- inputs: {"age_days": 2508, "days_push": 7, "days_rel": 7, "gap_med": 5.0, "n_releases_24m": 67}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 7245, forks 877 (observed 2026-08-28T04:09:57.799978+00:00)

## What it is
Flyte is an open-source workflow orchestration platform for coordinating data, ML models, and AI agents at scale, authored in pure Python and running on Kubernetes. It provides durable execution with retry, caching, checkpointing, and recovery from infra or code failures, plus a CLI and web UI.

## Use cases
- orchestrate machine learning training pipelines
- run durable AI agents with observability
- schedule and retry data processing workflows on kubernetes
- run high-throughput LLM inference jobs
- fine-tune models with distributed GPU compute
- build reproducible ETL pipelines in python
- recover long-running jobs from infrastructure failures

## When to choose
- you need production-grade ML/AI pipeline orchestration on Kubernetes
- you want durable, retryable workflows written in pure Python without YAML DSLs
- you need to coordinate agents, data, and model training in one platform
- you require checkpointing, caching, and failure recovery for long-running jobs

## When to avoid
- you only need simple cron-style job scheduling without orchestration
- your workloads are small single-machine scripts with no scaling needs
- you want a lightweight local-only task runner without a control plane
- you are not prepared to operate Kubernetes-based infrastructure

## Facets
- artifact type: framework
- maturity: active
- function: workflow-automation, scheduling, machine-learning, llm-inference, agent-framework, etl, deployment, cli, sdk
- domain: machine-learning, artificial-intelligence, developer-tools, cloud-computing
- platform: python, go, cli, cross-platform, cloud, self-hosted
- tags: workflow-orchestration, mlops, durable-execution, pipelines, kubernetes-native, declarative, grpc, fine-tuning, llm, ai-agents, lf-ai-foundation, data-engineering, containers, kubernetes, docker

## Member repositories
- flyteorg/flyte (main) score 95

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:09:57.799978+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 2026-08-29T17:39:44.225208+00:00, confidence not recorded.
  - readme: https://github.com/flyteorg/flyte (fetched 2026-08-28T04:09:57.799978+00:00, sha 37a6b79f9540)
  - homepage: https://flyte.org (fetched 2026-08-29T08:34:29.736029+00:00, sha a9ff3fb1607c)
  - site_page: https://docs.flyte.org/en/latest/api/flytekit/docs_index.html (fetched 2026-08-29T08:34:29.747367+00:00, sha 15da3dac3281)
  - site_page: https://flyte.org/integrations (fetched 2026-08-29T08:34:29.745355+00:00, sha 523a48ac05c2)
- Data as of 2026-08-30T08:39:29.467469+00:00.