Architecture | Mirora AI

01 PRODUCTION SERVICES

Live Infrastructure

Core microservices currently deployed and serving production traffic.

Core AI Service

LLM Gateway with multi-provider routing. Prompt Registry. Guardrails Engine. Metrics collection.

mirora-core-ai.onrender.com

Information Service

Schema-driven extraction. Gap analysis engine. Question generation. LLM-powered processing.

mirora-information.onrender.com

Document Processing

File ingestion (PDF, DOCX, images). Text extraction. Semantic chunking. Queue-based async jobs.

mirora-doc-processing.onrender.com

02 DEVELOPMENT SERVICES

In Progress

Services in active development, available for local testing.

Synthetic Data Service

Object, Event, Timeline, Arc, and Document engines for corpus generation.

localhost:8010-8015

Evaluation Framework

Ground-truth management. Automated eval runs. Regression detection.

localhost:8004

Vector Store Service

Unified interface to FAISS, Chroma, Weaviate. Collection management.

localhost:8009

03 INFRASTRUCTURE

Platform Stack

Compute

Render Web Services. Docker containers. Auto-deploy from GitHub.

Database

Supabase PostgreSQL. Transaction pooling. Managed backups.

Cache & Queue

Upstash Redis. TLS encryption. Serverless scaling.

LLM Provider

Groq API. Fast inference. Llama and Mixtral models.

04 SERVICE STATUS

Deployment Status

Service	Endpoint	Status	Dependencies
PRODUCTION
Core AI Service	api.mirora.app	✅ Live	PostgreSQL, Redis, Groq
Information Service	api.mirora.app	✅ Live	Core AI, Redis
Document Processing	api.mirora.app	✅ Live	Core AI, Redis
DEVELOPMENT
Synthetic Data Service	localhost	🟡 Dev	Core AI
Evaluation Framework	localhost	🟡 Dev	PostgreSQL, Redis
Vector Store Service	localhost	🟡 Dev	ChromaDB

05 DATA FLOW

Processing Pipelines

Document Ingestion

Upload → Document Processing → Text Extraction → Chunking → Information Service

End-to-end document processing with structured output

Career Analysis

Resume Upload → Document Processing → Information Extraction → Gap Analysis → Role Matching

Powers the Career Analysis tool

LLM Request

Request → Core AI Gateway → Provider Selection → Groq/OpenAI → Response Caching

Unified LLM access with caching and metrics

06 TECHNOLOGY

Platform Standards

Framework

FastAPI with async support. Pydantic models. OpenAPI docs. Structured logging.

Containerization

Docker multi-stage builds. Health checks. Render Blueprint deployment.

Security

API key auth (X-API-Key). CORS configuration. TLS everywhere.

Observability

Structured JSON logging. Health endpoints. Prometheus-compatible metrics.

Service Architecture