Overview
The SDLC Pilot is a flagship internal project exploring how AI can be deeply integrated into the software development lifecycle. It provides a web application and backend API that ingests Git repositories and enables AI-assisted workflows across the SDLC — from code review and documentation generation to test scaffolding and architectural analysis.
Technical Architecture
- Backend: Spring Boot application providing REST APIs for repository management, workflow orchestration, and result delivery
- LLM Inference: Local model inference via Ollama, enabling air-gapped operation and data sovereignty — critical for public sector contexts
- Job Queue & Workers: Asynchronous job queue architecture with dedicated workers for long-running AI tasks (code analysis, document generation, test synthesis)
- Frontend: Web interface for repository onboarding, workflow configuration, and result visualization
Key Design Decisions
- Local-first AI: All LLM inference runs locally via Ollama — no data leaves the network boundary. This was a non-negotiable requirement for the public sector context.
- Asynchronous processing: AI tasks are inherently long-running. The job queue architecture decouples request handling from processing, enabling scalability and resilience.
- Modular workflows: Each SDLC phase (review, test, document) is a pluggable workflow module, allowing incremental adoption.
Impact
The pilot demonstrated measurable acceleration in delivery throughput and is informing the broader AI strategy for the engagement.