Replicare
A decentralized platform to assess and improve the computational reproducibility of published empirical research
Built with
- React
- Node.js
- Express
- Firebase
- Material-UI
The problem
A large share of published empirical research cannot be computationally reproduced from its own stated methods and data — the code does not run, the data is missing, or the results do not regenerate. Assessing that at scale needs infrastructure, because doing it by hand for one paper is already a substantial effort.
How it was built
Designed and built the entire application: a platform for assessing and improving the computational reproducibility of published empirical research.
The architectural priority was performance and scalability, which is the right emphasis for this problem — reproducibility assessment is inherently work-heavy per paper, so the system has to handle that load without the interface degrading.
Built on React with a Node and Express backend and Firebase for data and authentication, a stack chosen to get a working platform in front of researchers quickly rather than to optimize for a scale it did not yet have.
Why this stack
- Firebase — authentication and data without standing up separate infrastructure early.
- Node and Express — the backend surface for assessment workflows.
Outcome
- Completed over roughly three months as the sole developer.
Want something like this built?
I build across Web3, AI, and financial infrastructure — smart contracts, data pipelines, LLM systems, and the interfaces on top.