This organization contains a cohesive set of AI backend microservices designed to mirror real-world backend and AI engineering systems.

The core focus is RAG architecture, supported by:

• embeddings (vector search building blocks)
• summarization (task-specific transformation)
• generative text (prompt-driven LLM output)
• optional orchestration (service coordination + evaluation)

All services are containerized, deployed, documented, and built with production patterns in mind.

flowchart LR

    Client --> RAG
    RAG --> GenAI
    RAG --> Summary
    RAG --> Client

    Embeddings
    Orchestrator

Notes: Embeddings and Summary are standalone microservices. GenAI is a standalone, prompt-driven generation service. Orchestrator is included as a supporting workflow layer.

A production-ready Retrieval-Augmented Generation backend built with Go, Postgres, and pgvector. Implements semantic search, top-K retrieval, and context preparation for downstream AI tasks.

📌 https://github.com/ai-backend-course/notes-memory-core-rag

The foundational CRUD backend for storing and retrieving notes. Demonstrates clean service layout, Postgres integration, logging, and metrics.

📌 https://github.com/ai-backend-course/notes-memory-core

A production-deployed prompt-driven Generative AI microservice built in Go. Exposes a clean /generate endpoint for text generation and acts as a reusable LLM capability for other backend services.

📌 https://github.com/ai-backend-course/ai-generative-service

An LLM-powered summarization microservice used to condense retrieved context or other text inputs. Designed as a task-specific service (summarization) rather than a general generation API.

📌 https://github.com/ai-backend-course/ai-summary-service

A dedicated embedding service supporting mock and real OpenAI embeddings, with validation, rate limiting, and observability.

📌 https://github.com/ai-backend-course/ai-embedding-microservice

A lightweight orchestration layer that coordinates retrieval + summarization, applies evaluation, and produces structured responses.

This service is not the core focus of the portfolio, but demonstrates how multiple AI backend services can be coordinated in a controlled workflow.

📌 https://github.com/ai-backend-course/agentic-workflow-service 🌐 https://agentic-workflow-service.fly.dev

A small Go backend running on AWS Lambda that persists data to DynamoDB and publishes asynchronous messages to SQS.

Designed to demonstrate core cloud-native backend fundamentals including stateless services, failure-aware design, clean separation of concerns, and AWS service integration.

📌 https://github.com/ai-backend-course/go-aws-backend-fundamentals

• Go (Golang) backend development
• Fiber v2: routing, middleware, handlers
• Retrieval-Augmented Generation (RAG) architecture
• Embeddings and vector search (pgvector)
• Service-to-service AI backend design
• Postgres 16 + pgxpool
• Semantic search and top-K retrieval
• Prompt-driven generative text APIs
• AI summarization pipelines
• Docker + multi-stage builds
• Fly.io production deployment
• TLS and container runtime configuration
• Logging, tracing, and observability
• Clean, production-oriented API design
• AWS Lambda (serverless execution)
• Amazon DynamoDB (NoSQL persistence)
• Amazon SQS (asynchronous messaging)
• AWS IAM (least-privilege service roles)
• Cloud-native, stateless backend design

This portfolio is centered on RAG-first AI backend systems, reflecting how many real-world AI applications are built today.

The projects emphasize:

• clear API boundaries
• production-ready backend patterns
• deployable services
• explainable AI pipelines
• incremental system design

The orchestration layer is included to show service coordination and workflow control, while the primary strength remains retrieval, data grounding, and backend reliability.

This work aligns with AI Backend Engineer, Backend Engineer (AI-enabled systems), and RAG-focused roles.

Jeff Ellis
AI Backend Developer
Email: jellis777@gmail.com
GitHub: https://github.com/jellis777