The Pre-Flight Readiness Gate
for India's AI Datasets
Help contributors answer the single most critical question before submission: “Is my dataset ready to be published on AI Kosh, and what must I fix first?”
Why Pre-Flight Tooling is the Missing Half of IndiaAI
The official AI Kosh MCP produces data quality reports only for datasets that have already been submitted, approved, and published. This creates significant friction: contributors (universities, research labs, state ministries) submit unstandardized dumps without knowing whether their schema, metadata, or privacy posture meets IndiaAI standards.
1. Verified Official Criteria
Evaluates candidate assets against the documented 3-section formula from real IndiaAI reports: Missing value density (15pts), Schema coverage (5pts), Usability (30pts), and Governance (35pts).
2. Sensitive Data Risk Signals
Executes UIDAI's mathematical Verhoeff Dihedral-10 checksum to detect valid Aadhaar number patterns, alongside Indian PAN and phone regex heuristics with clear advisory limitations.
3. Format Pragmatism
Respects public sector data realities. Never penalizes CSV formats. Evaluates schema readability and missingness fairly, offering optional Parquet conversion only where beneficial.
From Raw Data to Verified Submission
Contributor loads CSV or Parquet file in-memory. Zero data exfiltration to external servers.
Evaluates verified IndiaAI alignment score alongside extended column sparsity and duplicate checks.
Verhoeff Aadhaar & PAN regex scan flags potential identifier leaks with recommended masking rules.
Exports a verifiable pre-flight token and remediation checklist ready for MeitY AI Kosh intake.
Try a Live Contributor Pre-Flight Audit
Test clean agricultural benchmarks, Indic vernacular corpora, or sensitive sample records in real-time.