Alternatives for Modern AI Workflows
| SNo. | MS Tool/Service | OSS 1 (Rank) | OSS 2 | OSS 3 | OSS 4 | OSS 5 | GCP Equivalent | AWS Equivalent |
|---|---|---|---|---|---|---|---|---|
| 1 | Azure OpenAI Service | Hugging Face Transformers | Ollama | LlamaIndex | BricksLLM | FastChat | Vertex AI | Bedrock, SageMaker |
| 2 | Semantic Kernel | LangChain | Haystack | LlamaIndex | CrewAI | AutoGen | Vertex AI Agent Builder | Bedrock, Agents for Bedrock |
| 3 | Azure AI Foundry | TensorFlow Hub | Kubeflow | Hugging Face | PyTorch | FastAI | Vertex AI Workbench | SageMaker |
| 4 | Azure AI Evaluation SDK | Evidently AI | RAGAS | MLflow Evaluation | DeepEval, BLEU, ROUGE | OpenLMEval | Vertex AI Model Evaluation | SageMaker Model Monitor |
| 5 | Azure AI Inference SDK | DeepSpeed | vLLM | ONNX Runtime | Hugging Face Inference | TGI | Vertex AI | SageMaker |
| 6 | Azure AI Agent Service | AutoGen | CrewAI | FastAgency | Autogen | Langroid | Agent Builder (Vertex AI) | Agents for Bedrock |
| 7 | Azure Content Safety | Detoxify | Perspective API | WebPurify API | OpenAI Moderation API | Aestron | Content Safety API (Vertex AI) | AWS Content Moderation |
| 8 | Azure Language Service | spaCy | Hugging Face NLP | NLTK | Stanza | AllenNLP | Cloud Translation API | Comprehend, Translate |
| 9 | Azure AI Search | Meilisearch | Elasticsearch | Qdrant | Apache Solr | Lucene | Vertex AI Search | OpenSearch |
| 10 | Azure Document Intelligence | Tesseract OCR | EasyOCR | PaddleOCR | PaddleOCR | EasyOCR | Document AI (GCP) | Textract |
- Azure Functions
- Azure Kubernetes Service (AKS)
- Azure Container Apps
- Azure App Service
- Azure Logic Apps
- Azure Service Bus
- Azure Event Grid
- Azure Cosmos DB
- Azure Storage
- Azure Key Vault
- Azure Monitor
- Azure Active Directory
- Azure Redis Cache
- Dedicated Commeercial Vector DB:
- Pinecone - used by companies like MS, Shopify, Notion
- DataSax
- Zilliz cloud (managed milvus offering)
- Dedicated OSS Commercial Vector DB:
- Weaviate - hybrid search, uses GraphQL API
- Milvus - used by Salesforce, IKEA, Paypal
- Chroma - startups, smb
- ElasticSearch
- Qdrant - Bosch, Mozilla, Perplexity
- Faiss - Facebook AI similarity search
- Vespa - by yahoo, lexical search
- Deep Lake
- Vald
- Valkey
- Marqo
- LanceDB
- Traditiona databases with vector capabilities:
- PostgrSQL with pgVector
- OpenSearch
- Apache Cassandra
- ClickHouse
- Redis
- SingleStoreDB
- MongoDB Atlas Vector Search
- CockroachDB
- elasticsearch
=================================================================
- Azure Vector DB options
- Azure Cosmos DB for NoSQL - Native DiskANN indexing
- Azure Cosmos DB for MongoDB - vector search with embeddings
- Azure Cosmos for PostGreSQL - Native vector data type (preview)
- Azure Database for PostGreSQL - pgvector extension
- Azure Search AI - Built in vector search
- Azure SQL Database
- AWS
- Amazon S3 vectors
- Aurora postgreSQL - HNSW
- Amazon RDS for PostgreSQL - pgvector extension
- Amazon OpenSearch Service - Native k-NN
- Amazon Kendra - Built in intelligent search and vector capabilities
- Amazon MemoryDB in Redis
- Google
- AlloyDB for PostGreSQL - pgvector - scanNN indexing (scalable nearest neighbours)
- Cloud SQL for PostGreSQL - pgvector extension
- Vertex AI Vector search
- Firestore - vector embeddings + vertex ai integration
- Bigtable - vector embeddings + vertext ai integration
- Cloud Spanner - vectors columns wiht SQL extensions
=================================================================
- For text data embedding - create vector representations of words and sentences
- OpenAI’s text-embedding-3
- Google Gemini’s textembedding-gecko models
- Cohere’s Embed models
- SentenceTransformers
- FastText
- GloVe
- Word2Vec
- For image data - CNNs like VGG, Inception or CLIP from OpenAI - extract feature vectors from images
- For audio data - Spectrogram
=================================================================
Aug/19/2025
- Companies having universal API endpoint - abstraction framework to handle LLM's in the backend
- (1) OpenRouter - managed AI gateway focused on providing unified access to multiple LLM providers
- aggregates 400+ models from 50+ providers
- single API endpoint with OpenAI-compatible interface
- provider abstraction layer handles authentication, billing, and routing
- (2) TreuFactory - comprehensive MLOps platform that includes an AI gateway as part of a broader ML lifecycle management solution
- kubernetes-native platform deployed on your infrastructure
- (3) Microsoft AI Foundry Universal API Design
- easy model switching without code changes
- serves as the gateway to 1900+ models from multiple providers
- a single inference endpoint
- eliminates the need for developers to manage separate endpoints, authentication schemes, or API contracts for different model providers
- (4) LiteLLM - Multi-cloud flexibility, often cheaper, community-driven
- (1) OpenRouter - managed AI gateway focused on providing unified access to multiple LLM providers
=================================================================
Sep/20/2025
- Programming : Python, Jupyter labs, Hugging face, Pytorch, TensorFlow
- Monitoring: Streamlit, TensorBoard, Grafana
=================================================================
Sep/21/2025
- Agentic coding
- IDE Extensions - Github CoPilot, Cursor, Cline
- CLI toosl - Claude code, codex-cli, gemini-cli, aider, opencode
- Task specific tooks - coderabbit for code review
- vide-coding apps - lovable, bolt, v0
=================================================================