PASSMEMO — Structuring Human Narratives Through Data and AI

PASSMEMO is an AI-driven platform built for elderly care institutions, designed to capture, process, and transform personal life stories into structured digital narratives.

The platform bridges the gap between unstructured human input and usable, high-quality content through a combination of dynamic data modeling and processing pipelines.

The Challenge

The system had to handle highly variable and sensitive data sources:

• Life stories collected through interviews and dynamic forms
• Audio recordings requiring transcription and transformation
• Evolving data structures depending on context and user profiles
• Need for consistency despite non-standardized inputs

This required a system that is both flexible in input handling and strict in output structure.

The Approach

A backend-centric architecture was designed to support dynamic data ingestion and transformation.

At the core lies a dynamic form engine, allowing runtime definition of schemas without code changes. This enables:

• Adaptive questionnaires based on context
• Flexible data structures per use case
• Easy evolution of forms without redeployment

On top of this foundation, data processing pipelines were implemented to:

• Handle audio transcription workflows
• Normalize and clean heterogeneous data
• Map raw inputs into structured formats
• Prepare data for AI-driven narrative generation

From an architectural standpoint:

• Strong separation between data collection, processing, and output layers
• Emphasis on data normalization and schema abstraction
• API-first design for integration with frontend and content generation services
• Scalable pipeline design for handling increasing data volume

My Contribution

• Designed and implemented the full backend architecture
• Built the dynamic form engine and schema system
• Developed data ingestion and processing pipelines
• Structured and normalized complex, unstructured datasets
• Designed APIs for frontend and AI integration
• Ensured scalability, maintainability, and extensibility

Outcome

The platform enables institutions to:

• Capture rich and meaningful personal histories
• Convert unstructured input into structured, reusable data
• Generate readable and engaging narratives using AI
• Streamline data collection and processing workflows

Takeaways

Working with human-centered data introduces unique complexity.

This project reinforced that flexible data modeling, pipeline design, and separation of concerns are critical when transforming unstructured inputs into scalable, structured systems.