KIMI K3 NLP Integration
Budget: ₹12,500 – ₹37,500 INR
My current chatbot lives inside a React single-page web app and handles only rule-based replies. I want to replace that brittle logic with KIMI K3’s Natural Language Processing so users can type free-form questions and receive context-aware answers.
Scope
• Wire the KIMI K3 SDK or API into the existing React component that manages the chat window.
• Route every incoming user message through K3’s NLP engine, then return the parsed intent and generated response back to the UI without noticeable delay.
• Preserve my current session handling, user authentication and message history.
• Keep the code modular so that I can later toggle on K3’s voice recognition or speech synthesis if needed.
Tech details
The project is all browser-side React (no Next.js). State is managed with Redux Toolkit, while the backend is a simple Node/Express relay you can extend if the K3 API requires server-side secrets. The codebase is in a private GitHub repo; I’ll grant access once we agree on the approach.
Acceptance criteria
1. After integration, at least 90 % of a provided test set of user questions should be answered by K3 without fallback to the old rule system.
2. Average response time must stay under 1.2 s on a standard 4G connection.
3. All new code is pushed with clear commits and inline comments explaining K3-specific calls.
Deliverable: a pull request that integrates KIMI K3 NLP, an updated README explaining setup, and a brief video (or GIF) demoing the working chatbot in the browser.
Scope
• Wire the KIMI K3 SDK or API into the existing React component that manages the chat window.
• Route every incoming user message through K3’s NLP engine, then return the parsed intent and generated response back to the UI without noticeable delay.
• Preserve my current session handling, user authentication and message history.
• Keep the code modular so that I can later toggle on K3’s voice recognition or speech synthesis if needed.
Tech details
The project is all browser-side React (no Next.js). State is managed with Redux Toolkit, while the backend is a simple Node/Express relay you can extend if the K3 API requires server-side secrets. The codebase is in a private GitHub repo; I’ll grant access once we agree on the approach.
Acceptance criteria
1. After integration, at least 90 % of a provided test set of user questions should be answered by K3 without fallback to the old rule system.
2. Average response time must stay under 1.2 s on a standard 4G connection.
3. All new code is pushed with clear commits and inline comments explaining K3-specific calls.
Deliverable: a pull request that integrates KIMI K3 NLP, an updated README explaining setup, and a brief video (or GIF) demoing the working chatbot in the browser.