DevInsight

🛠️ How I built it

I used a Retrieval-Augmented Generation (RAG) pipeline powered by LangChain, Elasticsearch, and Vertex AI to analyze GitHub data.

• Backend: Flask-based API integrating GitHub’s REST API and our RAG engine.
  
• Frontend: React + Vite dashboard for real-time visual insights.
  
• Data Layer: BigQuery for analytics, Elasticsearch for semantic retrieval.
  
• AI Layer: Vertex AI models for contextual Q&A and summarization.
  

⚙️ Challenges I ran into

• Ensuring data consistency between GitHub’s live API and stored analytics.
  
• Handling complex queries in natural language and mapping them to GitHub data.
  
• Designing intuitive visualizations that convey insights without clutter.
  
• Managing RAG context windowing for large repositories efficiently.
  

🏆 Accomplishments that I'm proud of

• Built a working end-to-end AI insights system within hackathon time.
  
• Designed a clean, interactive dashboard with live GitHub metrics.
  
• Implemented semantic search + AI summarization for real-time Q&A.
  
• Created a vision that scales beyond GitHub — toward a full enterprise insight layer.
  

📚 What I learned

• How to design RAG pipelines that balance retrieval precision and generation speed.
  
• The importance of data visualization in conveying AI insights clearly.
  
• How collaboration and context-driven development can accelerate innovation.
  

🔮 What's next for DevInsight

While today DevInsight integrates with GitHub, the vision extends far beyond:

• Connect Jira, Slack, Confluence, Zendesk, Salesforce, and more.
  
• Build a unified intelligence layer for all enterprise data.
  
• Enable proactive insights — not just reports, but intelligent recommendations.
  

Built With

• bigquery
• elasticsearch
• fastapi
• langchain
• python
• react
• tailwindcss
• vertexai