~/s1re.sh $ ./audit --pipeline
I audit Nextflow and Snakemake pipelines for cost and wall time, then fix what I find. Rust when a step needs rewriting. Fixed fee, numbers before and after.
I will tell you within a day whether an audit is worth it.
01 — Engagements
One production pipeline, profiled on your data and infrastructure. A written cost and wall-time breakdown by process, a ranked list of fixes with the projected saving on each, and a 45-minute readout. No implementation.
The diagnostic plus implementation of the top fixes: resource right-sizing, spot and retry strategy, storage and staging layout, container and executor configuration. Before-and-after benchmarks on your own data and a handover document.
One bottleneck step reimplemented in Rust, benchmarked against the incumbent on your data, tested, containerised, and dropped into your existing workflow with nothing else changed.
Also available by the hour for Rust backend and performance work on existing services. Every proposal states what will be measured and how.
02 — How it goes
The execution trace and report from a recent run, or read access to your Seqera workspace. No code access needed for the diagnostic.
Within five business days: where the CPU-hours and dollars go, process by process, and the fixes ranked by saving per hour of work.
Implement with me on a fixed fee, or take the list to your own team. Either way you keep the report.
03 — Proof
Reads the execution trace and report that every Nextflow run already writes. Outputs per-process cost, allocation that was never used, what retries cost, and a right-sized config. Runs offline, on any executor.
github.com/OtoYuki/nf-auditLed an offline-first point-of-sale system from the Android device through the sync layer to the cloud backend, as Technical Director at press1-inc. In live testing in US stores.
More workContracts and invoices in USD, wire or Payoneer.