LiteParse v2, now in Rust 100x faster
Rust rewrite with PDFium delivers 100x speedup over the Python v1.
Sayiir: simple, embeddable durable workflow engine in Rust with Node.js & Python bindings. Checkpoint-based recovery, no deterministic replay. A simplified alternative to Temporal, Restate, and Airflow.
Checkpoint-based recovery beats determinism constraints; embed without separate server, unlike Temporal/Airflow.
Backend engineers, DevOps, data pipeline builders avoiding heavyweight orchestration
Temporal · Airflow · Step Functions
So I built something simpler and maybe i over simplified things..
It sticks with philosophy of "workflows as code", so no dsl , no no-code, or visual designer. just a developer friendly lib..
Sayiir is a library you pip/npm/cargo install. Define steps as normal async functions, it checkpoints each output. On failure, completed steps are skipped, execution picks up where it left off.
No replay, no determinism rules, no separate server to deploy, no activities — just your good old existing async functions with durable execution underneath.
Rust core, Python/Node.js bindings. MIT licensed
Docs: https://docs.sayiir.dev Playground: https://docs.sayiir.dev/playground/
Rust rewrite with PDFium delivers 100x speedup over the Python v1.
Continuation-passing durability with live/batch switching, but LangGraph and Temporal already own this space.
Durable workflows on Postgres tables beats Temporal and Inngest for zero-ops deployments.
L3 limit order book replay beats OHLC backtesting, but only matters if you're serious quant.
Remote LinuxCNC control via gRPC when native API only works locally.
Single Rust core with FFI bindings beats fragmented Python SDKs for consistency.