AI Business Analyst Intern
Genesis Networks Pte Ltd
- Built an end-to-end AI "document engine": four chained n8n workflows (research brief → client proposal → functional spec → test cases), each calling a self-hosted Qwen3.6-27B LLM, automating the early SDLC for client projects
- Exposed the workflows as tools under an n8n AI Agent via an MCP Server Trigger ("Project Assistant MCP Server"), with multi-file ingestion, a structured requirements register, and a webhook streaming documents from internal MinIO object storage
- Built and deployed a FastAPI tools/MCP microservice as the engine room for the tool logic, with its own repo, Dockerfile, Jenkins CI pipeline, and a Kubernetes Deployment and ClusterIP managed through Rancher, running fully in-cluster
- Stood up PostgreSQL in-cluster as the datastore, with tools writing structured records straight to Postgres
- Architected the system as n8n (MCP front door) → FastAPI (engine room) → PostgreSQL (store), alongside private in-cluster Qwen and MinIO services with no public endpoints
- Demonstrated the pipeline end-to-end on a sample (dummy-client) scenario: 22/22 functional requirements covered and 51 test cases generated, with bidirectional FR-ID coverage and no orphans
- Translated business requirements into technical implementation across the SDLC, working hands-on with RAG/embedding pipelines, GitLab, and Claude Code
Stack
- n8n
- MCP
- FastAPI
- PostgreSQL
- Qwen3
- MinIO
- Docker
- Kubernetes / Rancher
- Jenkins
- GitLab
- RAG
- Claude Code