Lakehouse BRD — Chapter 1: Overview

1.1 Strategic goals

The Lakehouse project is a core initiative in the organization’s digital transformation plan. Overall goals:

• Build a unified data platform serving multiple purposes: operations, reporting, risk management, partners, and AI/ML.
• Meet growing requirements for access speed, accuracy, and data quality control.
• Ensure legal compliance, in particular applicable personal data protection regulations.

Aligned with organization-level strategy, the Lakehouse system contributes to: better data-driven decisions; stronger credit and operational risk control; deeper digital partner integration and API adoption; and higher AI capability.

1.2 Project rationale

When data systems remain fragmented, common issues include:

• Slow data access: Manual aggregation of reports from many sources.
• Lack of standardization and lineage: Hard to trace data origin.
• Limited AI/ML scale: Models depend on manually aggregated data.
• Constrained partner integration: APIs cannot support real-time queries.
• Personal data compliance risk: Insufficient masking and access control.

1.3 Data ecosystem overview

Main component	Typical systems
CRM	In-house (may inherit existing)
Core Lending	Internal business applications
Payment	E-wallet, bank partner integration
Risk Engine	Python + SQL + dashboard
Reporting	Power BI + Excel rollups
DWH	MSSQL DWH, BigQuery (or equivalent)
API Layer	OpenAPI via Gateway
Logging / Audit	Distributed; needs consolidation

1.4 Lakehouse application scope

Subsystem / Business area	Application scope
Operations reporting	Multi-dimensional: by store, region, staff, product group
Financial analysis	Revenue, cost, profit, margin
Credit risk	Scoring, overdue debt, LTV analysis
Customer analysis	Lifecycle, behavior, target segments
AI & ML	Segmentation, risk prediction, up-sell
Partner integration	B2B API, controlled data sharing
Data control	Data masking, audit log, lineage, ownership

1.5 Definition of success

Criteria group	Success criteria
Technical	Process >10M records/day, ingest latency <5 min, query <3s
Business	Real-time operations dashboard covering all service points
ML & AI	Clean, complete data for at least 3 AI models in operations
Governance	100% of tables have a data steward and full metadata
Legal	Compliance with applicable data protection regulations and internal security standards