Sample Tech Stack 1: For a Large-Scale NLP Application with Knowledge Graph Integration on Azure
Context Representation and Storage
- Knowledge Graph: Azure Cosmos DB for Apache Gremlin
- Vector Embeddings:
- Azure Machine Learning Feature Store
- Consider Azure Virtual Machines or Azure Machine Learning Studio for open-source libraries (FAISS, Annoy).
- Explore Azure OpenAI Service (embeddings, vector search) or Azure Cognitive Search (vector search preview).
- Document Storage: Azure Cosmos DB or Azure Blob Storage with Azure OpenAI Service or Azure Cognitive Search.
- Relational Database: Azure SQL Database
Context Acquisition and Processing
- NLP Libraries: Python on Azure Virtual Machines, Azure Functions, or Azure Machine Learning Studio using spaCy, NLTK, Hugging Face Transformers. Consider Azure Cognitive Services for Language.
- API Integration: Python’s `requests` on Azure compute or Azure Functions. Consider Azure API Management.
- Stream Processing: Azure Event Hubs, Azure Stream Analytics, Azure HDInsight for Apache Kafka
Memory and Caching
- In-Memory Cache: Azure Cache for Redis
Infrastructure
Programming Languages
- Python, C#, Java, Go
Machine Learning Frameworks
- Azure Machine Learning (TensorFlow, PyTorch, scikit-learn)
Sample Tech Stack 2: For a Robotics Application Focusing on Environmental Context on Azure
Context Representation and Storage
- Spatial Data: Azure Spatial Anchors, Azure Cosmos DB (geospatial indexing), Azure Blob Storage
- Object Databases: Schemas within Azure Cosmos DB or Azure SQL Database
- Sensor Data Storage: Azure Time Series Insights, Azure Cosmos DB
Context Acquisition and Processing
- Sensor Data Ingestion: Azure IoT Hub, custom pipelines with Azure Functions or Azure Event Hubs & Azure Stream Analytics
- Computer Vision: Run OpenCV, PyTorch Vision, TensorFlow Vision on Azure Virtual Machines or use Azure Cognitive Services for Vision and Azure Custom Vision. Train models with Azure Machine Learning
- Sensor Fusion: Custom logic on Azure Virtual Machines or containers on AKS or ACI
Memory and State Management
Infrastructure
Programming Languages
- Python, C#, Java, Go
Machine Learning Frameworks
Sample Tech Stack 3: For a Dialogue System with Personalized Context on Azure
Context Representation & Storage
- User Profiles: Azure Cosmos DB or Azure SQL Database
- Dialogue State Tracking: Application logic on Azure Virtual Machines, Azure Functions, or ACI/ AKS with Azure Cache for Redis
- Conversation History: Azure Cosmos DB or Azure Blob Storage
Context Acquisition & Processing
- NLP Libraries: Python on Azure compute with Rasa (on AKS/ACI), spaCy, NLTK. Consider Azure Cognitive Services for Language and Azure Bot Service
- STT & TTS: Azure Cognitive Services for Speech to Text, Azure Cognitive Services for Text to Speech
- User Authentication: Azure Active Directory (Azure AD) or Azure AD B2C
Memory & Caching
Infrastructure
Programming Languages
- Python, C#, Java, Go
Machine Learning Frameworks
