How sepse/prishtina-open-brewery works
Project Type and Purpose
• This is a research-oriented, open-source hardware+software system (“Open Source Brewery”) for precision beer brewing.
• It combines:
• Embedded firmware on microcontrollers (ESP8266/ESP32/Arduino)
• Edge computing & data aggregation (Raspberry Pi + Python)
• Machine-learning recipe generation and analytics
• Web-based dashboards and user interfaces
• A public GitHub portal for code, hardware schematics, and datasets
Sketch the High-Level Architectural Layers
a. Hardware Layer
b. Firmware & Edge Layer
c. Data & Compute Layer
d. ML & Analytics Layer
e. Presentation & Integration Layer
f. Collaboration & External Services
Main Components & Their Roles
• Sensors (Temp, Humidity, CO₂, Flow)
– Physically attached to mash tuns, fermenters, and dispensing lines.
• Microcontrollers (ESP32/ESP8266/Arduino)
– Read sensor data.
– Control actuators (heaters, pumps, RFID locks).
– Publish data via MQTT/HTTP to the aggregator.
• Raspberry Pi (Edge Aggregator)
– Subscribes to sensor streams (MQTT broker or HTTP endpoints).
– Runs Python scripts to log data locally and forward to central DB.
– Hosts a local web dashboard (Flask/NodeJS).
• Central Database
– Time-series store (e.g., InfluxDB, TimescaleDB) for sensor logs.
– Relational or document store (e.g., PostgreSQL, MongoDB) for recipes and metadata.
• Machine-Learning / LLM Engine
– Batch jobs or microservice for recipe suggestion (e.g., Python + PyTorch).
– Data pipeline reads historical logs, trains/predicts, writes back recommendations.
– Optional REST/gRPC interface for on-demand queries.
• Web Dashboard & APIs
– Frontend (React/Vue) visualizing real-time data and ML suggestions.
– Backend API (FastAPI/Express) serving data from DB and ML engine.
• Open Resources Portal (GitHub)
– Source code (firmware/, backend/, ml/, docs/)
– Hardware schematics & BOMs
– Public datasets and community recipe contributions
Relationships and Data Flows
1. Sensors → Microcontrollers (analog/digital signals)
2. Microcontrollers → Edge Aggregator (MQTT/HTTP)
3. Edge Aggregator → Central Database (write data)
4. Central Database → Dashboards & Analytics (read data)
5. Central Database → ML Engine (training data)
6. ML Engine → Central Database (model outputs)
7. ML Engine → Frontend/API (recommendations)
8. User → Frontend/API (queries, recipe requests)
9. Contributors ↔ GitHub portal (code, data, docs)
Architectural Patterns
• Modular layers: hardware → firmware → edge → cloud → UI.
• Publish/subscribe (MQTT) between devices and aggregator.
• Microservices or functions for ML tasks.
• Time-series vs. relational DB separation.
• Community contributions via GitOps on GitHub.
Capturing Essential Details
• Raspberry Pi running both logging and local dashboard services.
• Distinguish time-series DB vs. relational DB for recipes.
• ML/LLM pipelines separately from real-time monitoring.
• User feedback loops (brewer queries ML, community submits recipes).