Design the Architecture of Pharmelis Registry (Global Pharmaceutical Data System) -- 3

We are building Pharmelis Registry — a canonical database for pharmaceuticals.

To make any pharmaceutical product understandable, anywhere, in any language.

Pharmaceutical data today is fragmented, inconsistent, and multilingual. There is no reliable way to identify and reconcile products globally across sources and markets.

Pharmelis Registry aims to capture real-world data and resolve it into clear, unambiguous product identities.

Objective

Design a system that:
- ingests heterogeneous pharmaceutical data (CSV, APIs, websites, PDFs, images)
- works across countries and languages
- handles messy, inconsistent data
- resolves product identity across sources
- produces a consistent global reference
- includes an internal dashboard to operate the system

The dashboard must allow:
- inspecting ingested data
- reviewing identity decisions
- monitoring system coverage
- adding and managing data sources

Deliverables

Provide a concise architecture document covering:
- identity resolution strategy (core part)
- ingestion approach for different source types
- core data model (what is stored vs computed)
- system architecture and data flow
- dashboard design and capabilities
- recommended tech stack and trade-offs

Required Profile

Strong in:
- Python
- data engineering / pipelines
- web scraping / ingestion
- API/backend development
- dashboard or web app development

Experience with messy data, search systems, or entity resolution is expected.