WORK / 04 · Materials-science research environment

Mendeleev

Materials research needs a substrate, not another disconnected tool.

Mendeleev is an India-first working environment for materials science. It brings literature, extracted properties, crystal candidates, computation, researcher discovery and patent prior art into one continuing enquiry so that evidence remains attached to the decisions made from it.

Live v0 · development2026
A cobalt materials-science observatory organised around a faceted crystal-analysis instrument
Materials observatory / literature, structures, computation and people held in one field
Evidence position

A September 2026 project record documents a live v0 and reports its corpus, tools and architecture; several production and evaluation claims remain incomplete.

Condition
Live v0 · development
Review
Foundation source and reproducibility review / active
Record year
2026

THE GOVERNING QUESTION

What changes when a researcher can move from an idea to literature, candidate materials, computation, collaborators and prior art without leaving one evidence-bearing environment?

THE INTERVENTION

What the work changes.

  1. 01

    Join fragmented research surfaces without concealing the provenance of the evidence moving between them.

  2. 02

    Make Indian materials literature and researcher networks first-class objects rather than residual filters inside a global interface.

  3. 03

    Bring crystal candidates, property prediction and visual scientific instruments into the same conversational working context.

  4. 04

    Treat patent novelty as a decomposed evidentiary question across composition, structure, method, application and property—not as a single confidence score.

A circular materials evidence table connecting crystal specimens, papers, charts and researcher networks
Evidence ring / composition, structure, literature, computation, people and prior art

ANATOMY OF THE WORK

One proposition, held through several layers.

01

Literature and data

Indian-affiliated materials papers, normalized property records and source spans supporting traceable retrieval.

02

Computation and discovery

Crystal candidates, scientific Python, property prediction and visual instruments for comparing material systems.

03

People and lineage

Researcher, collaborator and advisor-lineage graphs designed to make distributed Indian capability more legible.

04

Novelty and prior art

Facet-based patent search joined to the closest literature, time, geography and assignee evidence.

CLAIM REGISTER

What may—and may not—be said.

Each material claim is paired with the present basis on which it rests. Absence of evidence is kept visible rather than converted into confidence by presentation.

C/01

A public Mendeleev v0 exists.

The September 2026 project overview records mendeleev.in as live from 15 May 2026.

Documented
C/02

The working corpus and researcher graph have reached the reported scale.

Counts are a dated project snapshot and have not yet been independently reproduced by the Foundation.

Documented
C/03

Patent-search functions and the novelty orchestrator have been built and tested.

The project record reports backend verification while copilot wiring and final frontend rendering remained incomplete.

Partial
C/04

The environment improves scientific discovery or institutional outcomes.

No benchmark, controlled evaluation or institutional outcome record is supplied.

Not established

RESEARCH RECORD

The objects presently in the file.

R/01Public system

Mendeleev v0

A public India-first materials-research environment is identified in the September project record.

Documented live
R/02Corpus record

Literature, properties and structures

Dated corpus counts and the extraction architecture are recorded but not yet independently reproduced.

Source-reported
R/03Research instrument

Scientific and collaboration tools

Sixteen tools and five visual renders are described across materials analysis and researcher discovery.

Implemented in part
R/04Development boundary

Authoring, education and novelty interface

These layers were planned or only partially wired at the date of the source record.

Incomplete

WHAT EXISTS NOW

The present record.

  • A public v0 is reported live at mendeleev.in from 15 May 2026.
  • The September project record reports 17,183 papers, 15,749 property rows, 1.24 million crystal candidates, 10,000 researchers, 1.33 million collaboration edges and 99,990 advisor-lineage records.
  • Sixteen purpose-built research tools and five rich visual renders are described in the operating record.
  • The record describes an India-filtered literature pipeline with source-span provenance and a Mumbai-hosted application stack.

WHAT IS NOT YET CLAIMED

The boundary of the evidence.

  • Corpus and graph counts are attributed to a dated project record; they are not presented as a completed independent audit.
  • Extraction accuracy, property normalization, predictive uncertainty and source-span fidelity require published evaluation.
  • Authoring and education are planned layers, not current operating claims.
  • Market estimates, pricing, fundraising and growth targets in the project deck are strategy—not Foundation findings.

WHY IT BELONGS TO THE FOUNDATION

Mendeleev asks whether a scientific institution can possess more than access to papers and models: whether it can maintain a connected substrate of evidence, computation, people and prior art through which national research capability becomes easier to find, test and extend.

NEXT PASSAGE

What must happen before the work can claim more.

  1. 01

    Publish a dated methods and corpus note with update cadence, filters and reproducible samples.

  2. 02

    Evaluate extraction accuracy, provenance fidelity and computational uncertainty.

  3. 03

    Complete and document the novelty-analysis interface without overstating legal conclusions.

  4. 04

    Establish the institutional relationship, governance and long-term operating responsibility of the platform.