Technology Stack

An overview of the technologies, cloud services, and tooling that power the Zynomi platform.

Deployment Architecture

Architecture Philosophy

Zynomi is built on a simple, modern technology stack. While there are multiple components, each serves a specific purpose in delivering a comprehensive clinical trial platform.

Principle	Description
Loosely Coupled	Components are independent and can be updated separately
Plugin-Based	Features can be added without modifying core code
Horizontal Scaling	New capabilities added alongside existing ones
Microfrontend Ready	AI Chatbot installs as standalone widget

Cloud Infrastructure

Core Services

The platform is delivered as a self-contained Docker Compose bundle on AWS EC2 — one dedicated stack per client, in the AWS region the client selects.

Service	Provider	Purpose	Notes
Web App and Mobile Website	Docker Compose	Next.js hosting, REST APIs	Self-hosted on AWS EC2
Healthcare Backend	Frappe (self-hosted)	EHR, patient data, API backend	Docker Compose, part of the bundle
Database	Supabase (self-hosted)	PostgreSQL, authentication, real-time	Docker Compose, part of the bundle
Cache Store	Redis	Cube.dev cache, session management	In-memory for small deployments
Source Control	GitHub	Code repository, CI/CD	Free tier available

Mobile push notifications

Push delivery to the Sublink mobile app uses Firebase Cloud Messaging (FCM) — the standard delivery channel for Android/iOS push. It is an optional, app-level messaging service; all platform hosting runs on AWS.

Analytics Infrastructure

Service	Provider	Purpose	Notes
Data Ingestion	dlthub (Python)	OLTP to lakehouse pipeline	Python-based ETL
Data Transformation	dbt	Medallion architecture, data marts	OSS Core edition; Cloud available for scaling
Semantic Layer	Cube.dev	Governed metrics, REST/SQL APIs	OSS edition via Docker Compose; Cloud available for enterprise
Data Lake	PostgreSQL (Supabase)	Default for small deployments	Can swap to Snowflake, Iceberg, Databricks
Data Lineage	Marquez (OpenLineage)	Pipeline lineage tracking and visualization	Web UI at `http://localhost:3300`

AI and Integration Services

Service	Provider	Purpose	Notes
API Gateway	KrakenD	Unified API endpoint, rate limiting	Self-hosted via Docker Compose
MCP Server	Custom Built	Semantic Layer as MCP tools	Adheres to Cube.dev API/SDK standards
AI Agent / Chatbot	Custom Built	Agentic analytics interface	Not Cube.dev Cloud (enterprise-only)
LLM Provider	OpenAI (Default)	GPT-4.0 Mini for tool selection	Any commercial or OSS LLM supported

LLM Requirements

An LLM API subscription is required for the AI chatbot. Default is GPT-4.x Mini. Better models improve tool selection accuracy. No hallucination risk as responses are grounded via MCP tool-calling.

Mobile Application

Platform	Technology	Description
iOS	Native Shell	WebView wrapper loading Ionic mobile website
Android	Native Shell	WebView wrapper loading Ionic mobile website
Mobile Website	Ionic + Vue.js	Progressive web app hosted via Docker Compose

Security and Compliance

The platform is designed to support regulated clinical workloads. Compliance is a shared, ongoing program — the controls below describe what the architecture provides today; formal certifications and audits are on the roadmap, not claimed as delivered.

Control	Description
Audit Trail	Activity logging across services (a visual audit-trail UI for 21 CFR Part 11 / GxP workflows is on the roadmap)
Data at Rest	Encrypted storage for databases and files
Data in Transit	TLS 1.3 for all communications
HIPAA	Architecture designed to support HIPAA-aligned deployments (dedicated per-client stack, encryption, access control)
GDPR	Architecture designed to support GDPR-aligned data handling (client-selected AWS region, data isolation)

Development Environment

Hardware Requirements

Specification	Minimum	Recommended
RAM	8 GB	16 GB
CPU	4 cores	8 cores
Storage	50 GB SSD	100 GB SSD