Skip to content
View VivienP's full-sized avatar

Block or report VivienP

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
VivienP/README.md

Vivien Perrelle

Scientific Software Engineer — AI for Science

Building reliable software at the boundary between models, scientific data and experiments.

Models → Decisions → Experiments → Data → Verification → Next experiment ↺

Website · LinkedIn

Selected work

Which experiment should we run next?

Epibudget — Experimental design for protein engineering under limited laboratory budgets. I benchmark strategies for selecting which protein variants to measure when only a small fraction of the sequence space can be tested.

Protein engineering · Experimental design · ESM-2 · Evaluation

Can experimental data remain traceable and reproducible?

LabBridge — Provenance-first infrastructure for experimental science, turning scientific files into immutable source artifacts, versioned experiments, traceable transformations and independently verifiable packages.

Scientific data · Provenance · PostgreSQL · MinIO · Reliability

Can scientific outputs be verified?

LocusLab — Infrastructure for inspectable and reproducible scientific evidence workflows. Related: Scientific Claim Verification Engine — deterministic verification of scientific claims against their supporting evidence.

Scientific AI · Verification · Reproducibility · LLM evaluation

From lab to software

My interest in reliable scientific software started with physical systems.

I built BioWatch — a smartwatch prototype combining electrochemical biosensors, embedded electronics and software — and later worked on in-vitro testing of microneedle glucose-monitoring prototypes at PKvitality.

That experience shaped how I think about AI for Science: a useful model is only one part of a reliable scientific system.

Interests

Experimental design · Scientific data provenance · Lab automation · Scientific agents · Closed-loop AI for science

Latest notes

Epistasis Explained: Why the Best Protein Variant Is Not Always the Best Experiment August 13, 2026 · 12 min

Measure for Information, Not for Fitness: Designing Protein Experiments to Reveal Epistasis July 23, 2026 · 15 min

Currently

I'm interested in engineering roles or freelance work building software close to scientific models, experimental systems and closed-loop biology/chemistry R&D workflows — particularly scientific platforms, reliability and lab automation.

Pinned Loading

  1. epistasis-budget epistasis-budget Public

    Experimental-design methods for budgeted protein-epistasis mapping with zero-shot ESM-2 signals; evaluated on GB1.

    Python 2

  2. scientific-claim-verification-engine scientific-claim-verification-engine Public

    Python 3

  3. locuslab locuslab Public

    Local CLI that verifies exported MDR/IVDR dossiers and writes an Evidence Trace Audit package.

    Python

  4. labbridge labbridge Public

    Provenance-first experimental-data platform and fault-aware campaign runtime for electrochemical R&D. Status: specification, no implementation yet.

    Python