AI KOSH Product Strategy Dossier
Comprehensive policy, technical, and architectural strategy documents for IndiaAI Mission at the Ministry of Electronics & IT (MeitY).
AI KOSH Product Strategy: Pre-Flight Contributor Readiness Layer
Ministry of Electronics and Information Technology (MeitY) • IndiaAI Mission
1. Problem Statement: The Pre-Flight Contributor Gap
AI KOSH currently hosts over 16,000 datasets. However, live MCP tools (get_dataset_data_quality_report_tool) reveal that data quality scoring happens downstream — only after a dataset is already ingested and published.
This creates a major operational bottleneck: contributors (researchers, state ministries, startup consortia) submit unvalidated dumps without knowing whether their schema, metadata, or privacy posture meets IndiaAI standards. By providing a client-side pre-flight verifier, contributors answer the question: “Is my dataset ready to be published on AI Kosh, and what must I fix first?”
2. Strategic OKRs for the Fellow Mandate
Self-serve automated pre-flight scoring eliminates review backlogs.
UIDAI Verhoeff checks detect Aadhaar digits before files leave the contributor machine.
High-quality 4.5+ star verified assets unlock IndiaAI GPU compute vouchers.
3. Architectural Principles
- Official IndiaAI Rubric vs Extended Checks: Clear distinction between verified government metrics (Hygiene, Usability, Governance from actual reports) and heuristic diagnostics (duplicate rates, column sparsity).
- Technical Risk Signals vs Legal DPDP Compliance: Algorithmic screening detects potential Aadhaar and PAN leaks, with explicit disclaimers that statutory DPDP Act 2023 compliance remains the contributor's legal accountability.
- Format Pragmatism: Never penalizes standard CSV/JSONL formats; offers optional Parquet conversion only for massive datasets (>50k rows).