100x More monthly users
3 seconds From request to graphed data
200% More insights generated
Harmonized UI for both simple graphs and complicated analyses
The challenge
Our client, a major International Bank, needed to make its vast financial data repository accessible to a broader audience internally.
Their existing data builder was unintuitive and limited users to basic charts, alienating non-technical users. For technical users, it lacked the ability to conduct deeper analyses or combine data from multiple sources.
PortBlueSky developers are pivotal during tech transitions and the need to scale. Their proactive leadership keeps us on schedule and their expertise consistently exceeds our quality expectations.
Our solution
PortBlueSky led the development of a 'virtual data scientist’. Users could simply ask questions, and the AI would generate insights, graphs, and tables, trans-forming complex queries into simple conversations.
The virtual assistant handled pulling data from various sources, checking access rights, running analyses, providing options, and graphing & interpreting results.
A dedicated PortBlueSky team of AI specialists and developers handled the project from end to end.
Technologies
Backend: Node (TypeScript, next.js), Python (fastapi), Postgres, Redis, TimeScaleDB
Frontend: React (TypeScript, next.js, Hooks), Playwright, Storybook, Tailwind
AI / Machine Learning: OpenAI GPT, LLaMa 3, Scikit, PyTorch, Peft, Huggingface, Azure OpenAI Service
Automation: Kubernetes, Docker, Helm, GitHub Actions
