• AI backend services with clear APIs, durable state, and human-in-the-loop flows
• Agentic workflows using LangGraph, LangChain, tool calling, and structured outputs
• RAG and knowledge systems for document ingestion, retrieval, and answer generation
• FastAPI, Django, and NestJS backends backed by PostgreSQL and Dockerized services
• Workflow automation and data pipelines with n8n, RabbitMQ, object storage, and Python data tooling

I care about the parts that make AI products useful outside a demo: validation, state management, evaluation, observability, failure modes, security boundaries, and maintainable architecture.

AI Backend Engineering
Agentic Workflow Design
RAG APIs and Knowledge Pipelines
Production FastAPI Architecture
Local-first LLM Systems
Human-in-the-Loop AI Products
Backend Reliability and Evaluation

Local-first AI recruiting decision support built with FastAPI, PostgreSQL, LangGraph, LM Studio, and Next.js.

Repository: AliEslah/recruitment-agent

Highlights:

• FastAPI backend with SQLAlchemy, Alembic, RBAC, audit logs, and redaction boundaries
• LangGraph workflows for job calibration, candidate scoring, interview planning, and evaluation
• Local LLM execution through LM Studio's OpenAI-compatible API
• Human review checkpoints for shortlist and final hiring decisions
• Docker Compose local stack with PostgreSQL, Mailpit, backend service, and frontend
• Backend and frontend test coverage with documented validation commands

• Clear module boundaries and API contracts
• Typed data models, validation, and explicit failure modes
• Product workflows designed around human review and operational constraints
• Local development that is reproducible with Docker, documented commands, and tests
• AI systems built with evaluation, observability, and safety boundaries in mind

I am sharpening this GitHub profile around AI backend engineering. The next portfolio-grade repositories I plan to build or publish are:

• rag-knowledge-api: document ingestion, retrieval, citations, and evaluation for a production-style RAG API
• fastapi-production-template: a reusable backend template with auth, settings, migrations, CI, Docker, and observability
• langgraph-agent-workflows: practical agent workflow patterns for approval gates, retries, state, and tool execution
• n8n-llm-agent-nodes: safe workflow automation examples for LLM-backed operations, if no private/company code is involved

Building backend systems for reliable AI products.