Team
kairos
Project Concept
No description has been added yet.
Entry
Status: Submitted
Last saved: December 11 at 9:30 PM NZDT
Team Roster
You must be registered for the event to view the team message board.
Jason Xavier Team Lead RSVP Approved
Fullstack Software Engineer at SmartTrade
# ElevenLabs Team Contributions
## Rohan Drummond + Sean Outhwaite
**Role:** Voice Integration Developer / Frontend Engineer
**Contributions:**
- Integrated **ElevenLabs React SDK** (`@elevenlabs/react` v0.12.1) for voice conversation functionality
- Built complete voice chat component (`apps/ui/app/routes/elevenlabs/voice-chat.tsx`) with real-time audio streaming and bidirectional communication
- Created ElevenLabs-specific UI component library:
- **Orb visualization component** - 3D animated orb using Three.js and React Three Fiber that responds to agent state (thinking, listening, talking) and audio volume
- **Conversation UI components** - Full conversation interface with scroll areas, message rendering, and empty states
- **Message components** - Styled message display with role-based rendering (user/assistant)
- **Response components** - Voice response handling and display
- **Shimmering text component** - Animated text effects for loading states
- **Avatar, button, card, scroll-area components** - Styled components specifically designed for voice interface
- Implemented API route integration (`apps/ui/app/routes/api.elevenlabs/route.ts`) for ElevenLabs backend communication
- Configured ElevenLabs Agent ID integration via environment variables (`ELEVENLABS_AGENT_ID`)
- Added voice mode toggle functionality and microphone stream handling with proper error handling
- Implemented bidirectional voice communication:
- **Speech-to-Text** - Real-time voice input processing using ElevenLabs conversation API
- **Text-to-Speech** - Voice response generation with natural-sounding audio output
- Integrated `useConversation` hook from ElevenLabs React SDK for managing conversation state, connection lifecycle, and message handling
- Added comprehensive installation and setup instructions to README for hackathon reproducibility
- Adapted API routes specifically for hackathon project requirements
**Specific Sponsor Tools & APIs Used:**
- **ElevenLabs React SDK** (`@elevenlabs/react` v0.12.1) - Voice conversation hooks (`useConversation`) and React components
- **ElevenLabs Agent API** - Conversational AI agent integration for voice-based interactions
- **ElevenLabs Speech-to-Text API** - Real-time voice input processing and transcription
- **ElevenLabs Text-to-Speech API** - Natural-sounding voice response generation
- **ElevenLabs Conversation API** - Bidirectional voice streaming and conversation management
I’m Jason, a full-stack engineer at SmartTrade. I mainly work on internal tools and workflow automation across our development platform. Recently I’ve been building an MCP tool to help automate things. I enjoy practical problems, and I’ve always had a genuine love for both learning and teaching.
I’m pretty active across the local AI/engineering meetups, including Prompt Poets Society and JuniorDev Auckland.
Always keen to talk AI and practical engineering. Mainly looking to connect with founders and senior engineers who are building things.
I’ve been building an MCP server that exposes our enterprise development API to AI tooling. It replaces our in-house IDE and several specialized workflows with a fully Cursor-compatible setup. The server handles authentication, runs production/test DB transactions, pulls customer ticket data, queries an aggregated docs source, can validate code against archives, can executes tests, and supports a wide range of internal searches.
Richard Fontein RSVP Approved
Founder at TauCAD
# Tau Team Contributions
## Richard Fontein + Jason Xavier
**Role:** Platform Developer / Full-Stack Engineer
**Contributions:**
- Maintained and improved existing Tau platform infrastructure to support voice integration
- Fixed file tree editor operations for better UX during voice interactions, ensuring seamless file management while using voice commands
- Enhanced markdown rendering with Streamdown library for improved message display in chat interface
- Improved chat message reasoning collapsible behavior for better information organization
- Implemented automatic textarea focus after new chat messages to improve user experience flow
- Fixed file explorer machine path handling to ensure correct file operations during voice-driven workflows
- Optimized error stack trace visual design for better debugging and user feedback
- Adjusted code viewer syntax highlighting delay for smoother rendering performance
- Built memory optimizations for UI build process to support additional voice features
- Maintained compatibility with existing CAD kernels (Replicad, OpenSCAD, JSCAD, KCL) throughout voice integration
- Ensured platform stability and reliability for voice integration by fixing edge cases and improving error handling
- Updated build configurations and dependencies to support new voice features
**Specific Technologies & APIs Used:**
- **LangGraph** - Multi-agent orchestration system with supervisor pattern coordinating research expert and CAD expert agents
- **LangChain** - Tool integration framework and LLM abstraction layer for agent reasoning
- **Tavily API** - Web search integration for research agent to fetch reference materials and specifications
- **OpenAI API** - LLM provider for agent reasoning and code generation (GPT-4o-mini, GPT-4o)
- **Anthropic API** - Alternative LLM provider support for agent reasoning (Claude models)
- **WebSocket** - Real-time bidirectional communication infrastructure for chat and agent updates
- **React 19** - Frontend framework with latest features for UI components
- **NestJS** - Backend API framework with Fastify adapter for high-performance request handling
- **PostgreSQL with Drizzle ORM** - Database layer for conversation history and agent state persistence
- **XState V5** - State machine orchestration for complex UI workflows
- **Three.js & React Three Fiber** - 3D rendering for CAD model visualization
I'm Richard, founder of TauCAD, the open-source computer aided design (CAD) framework for the web (https://tau.new, https://github.com/taucad/tau).
Tau enables anyone to design 3D items with natural language, whilst providing Typescript libraries for developers to integrate CAD into their web app. My vision for Tau is to enable anyone to build robots with code using code-based CAD for 3D models, firmware and AI integrations.
I'm a Product Engineer, Founder, and Builder. I am passionate about open-source software and LLMs and love building tools for developers.
Before founding TauCAD, I most recently worked at Novu (https://novu.co, https://github.com/novuhq/novu) developing notification infrastructure & frameworks, where we delivered notifications to over 200 million people every month.
- Mentorship in building AI-chat experiences.
- Seeking pre-seed/seed investors
- Connect with potential hires interested in developer tooling & computer aided design
I am currently building TauCAD, the open-source CAD (computer aided design) platform for the web.
Tau exists to make CAD accessible to everyone, empowering creators to access world-class CAD software from their browser.
Tau features AI-chat based editing in a Cursor-style editor built specifically for CAD.
Github: https://github.com/taucad/tau
Website: https://tau.new