Full-stack engineer
Tarka Sabha
Tarka Sabha is a multi-agent AI debate platform that coordinates prompts, provider access, generated arguments, and debate-state flows.
Workflow proof
Debate workflow state
The product separates debate setup, agent turns, provider calls, and generated output.
- Topic setup
- Agent roles
- Provider boundary
- Debate output
Proof
- Backend boundary keeps provider credentials out of the browser.
- Debate workflow state is separated from provider-specific logic.
Problem
AI debate workflows need a structured way to coordinate multiple model providers, user prompts, and credentials without mixing product logic with provider-specific details.
Users and context
Built for users who want to run and compare AI-assisted debate flows through a web interface.
Product workflow
A user defines the debate context, configures model access, starts the debate flow, and reviews generated arguments from participating AI agents.
Architecture
React powers the client experience, while a Node.js backend coordinates provider requests, credential handling, and debate workflow state.
Key engineering decisions
- Separated provider orchestration from the interface so model integrations can change without reshaping the user flow.
- Kept credential handling behind backend boundaries instead of exposing provider keys to the browser.
- Structured the project around the debate workflow rather than scattered prompt experiments.
Tradeoffs and limitations
- Supporting multiple providers increases orchestration complexity compared with a single-model implementation.
Next improvements
- Add richer debate history views.
- Improve response comparison and citation displays.