ML2Know
How to…
A secure, governed Data Lake that unifies all corporate knowledge and becomes the foundation for a knowledge base queryable in natural language: company documents made searchable with RAG, always respecting roles and data sensitivity.
Medallion Architecture
Three-layer storage — Bronze (immutable raw), Silver (curated and normalized), Gold (ready to consume) — with tracked transformations from heterogeneous sources.
Governance & Classification
Every data point labeled by type (public, internal, confidential, personal) and security level: two orthogonal dimensions driving access and protection.
Roles & Access
Role-data mapping (Data Engineer, Analyst, Scientist, Business User, Audit, Admin) with effective access defined by the roles × classification matrix.
Access Control
RBAC and ABAC, data policies (row-level security, dynamic PII masking) and full audit: control travels with the data, not with the application.
Knowledge Base & RAG
On top of the lake, vector indexing of documents and retrieval-augmented generation to answer “how to…” citing sources, with the same permissions as the lake.
Continuous Security
End-to-end encryption in transit and at rest, data lifecycle (retention and deletion) and active monitoring to detect anomalous access.
From data to answer: a single, secure and governed data foundation that feeds analytics, BI, machine learning and the conversational assistant over corporate knowledge, always with tracked, role-consistent access.