evyAI
Fullstack AI Developer · Contract · Remote
Rebuilding an AI networking assistant: 50x lower Firestore costs and a 10x smaller bundle
I lead the development team that took evyAI from a successful MVP to a production platform, cutting costs and bugs while improving performance.
- lower Firestore bill
- 50x
- smaller frontend bundle
- 10x
- automated tests and deploys
- CI/CD
The context
evyAI is an AI assistant for professional networking. It ships as a React browser extension that works inside dynamic web pages, backed by a Node.js and Firebase platform with paid subscription tiers. After the MVP proved the product, the codebase needed to scale without the costs scaling with it.
What I did
- Led the team that revamped and refactored the platform after a successful MVP, reducing costs and bugs while increasing performance.
- Refactored the critical paths and cut the Firestore bill by 50x and the bundle size by 10x, while starting a new codebase that follows best practices.
- Architected AI workflows with LangChain for persona-based content generation and context-aware networking tools.
- Built a low-latency React browser extension that gives users real-time AI assistance inside dynamic web pages.
- Built and optimized a Node.js API layer for high-concurrency LLM requests and real-time data processing.
- Integrated and maintain Stripe billing with multi-tier subscriptions and global payments.
- Added CI/CD with GitHub Actions and a full test environment for the frontend and backend repos.
- Integrated Slack, ActiveCampaign and HelpCrunch so the business reaches more leads and answers support faster.