NexaGrow

NexaGrow pairs ClickHouse OLAP with Gemini AI to turn millions of raw film metrics into instant box office decisions in milliseconds.

Comment

🎬 Inspiration

In the fast-paced film and entertainment industry, studio executives and production crews are drowning in millions of rows of data—including budget figures, box office returns, demographic records, and viewer sentiments.

Traditional database systems are simply too slow for live analytical queries, and standard dashboards lack the flexibility to answer unique, custom questions. We wanted to build a zero-friction, incredibly fast, and smart tool that puts instantaneous blockbuster insights right at a decision-maker's fingertips. That's why we created NexaGrow.

🧠 What it does

NexaGrow is an autonomous, full-stack Studio Intelligence Hub that pairs the ultra-fast analytical power of ClickHouse OLAP with the cognitive automation of Gemini Agentic models.

Key features include:

  • Natural Language to ClickHouse SQL: Users can chat with our AI in plain English (e.g., "Which film genres had the highest ROI with budget-to-gross margins?"). Gemini translates this into highly optimized SQL, runs it against millions of database rows, and returns the parsed results instantly.
  • Interactive SQL Playroom: A rich database sandbox pre-loaded with blockbuster movie tables to let users query and run SQL code directly in their browser.
  • Studio Metrics Dashboard: Live charting of box office ROI, production budget tracks, and audience sentiment analyses using fluid Recharts visualizations.
  • TechOps & Delivery Console: A command center showing system latency, query duration metrics, and API health to ensure production-level performance.

🏗️ How we built it

We engineered NexaGrow using a robust, desktop-first responsive stack:

  • Frontend: Built on React 18 and Vite for fast load times, styled with custom warm-neutral Tailwind CSS classes, and animated with Framer Motion.
  • Visualizations: Rendered with high-fidelity vector graphics and charts powered by Recharts and Lucide.
  • Backend: Powered by an Express.js server that handles database connections, proxy calls, and secure Gemini interactions.
  • OLAP and AI Pipeline: Integrated the ClickHouse database engine with the newest official @google/genai SDK to run lightning-fast analytics.
  • Deployment: Hosted on Render with full continuous integration, custom public-root vector favicon assets, and automated building pipelines.

⚡ Challenges we faced

Grounding a Generative AI agent to write syntactically correct SQL queries for a columnar database without hallucinations was a major challenge. We solved this with advanced prompt-engineering constraints, providing schema maps, and setting up rigorous error-handling pipelines.

Additionally, ensuring that our high-fidelity custom branding vector assets compiled cleanly in Vite on low-latency deployed container servers was tricky, but we resolved it by implementing a public-root asset serving pipeline.

🏆 Accomplishments that we're proud of

  • Sub-Second Analytics: Generating complex database query responses from millions of mock data records in milliseconds.
  • Pixel-Perfect Fluid UI: Designing a clean, immersive dark-luxury dark mode dashboard that is responsive and highly intuitive.
  • Full Playground Sandbox: Building an entirely self-contained database simulator fallback so judges can query and play with data out-of-the-box with zero setup required.

📚 What we learned

We learned how to deeply couple AI workflows with columnar OLAP architectures (ClickHouse) for real-time analytics. We also mastered state sync patterns between client-side dashboards and server-side model pipelines.

🚀 What's next for NexaGrow

We plan to introduce advanced machine learning predictive forecasting for opening-weekend box office numbers, and integrate the Gemini Live API for real-time voice-interactive studio meetings.

Built With

Share this project:

Updates