
Gordon Ramsay Chatbot
Persona-Driven Cooking Assistant with LLM Prompt Engineering
“A useful chatbot needs both personality control and reliable response structure.”
System Design
This web chatbot is built with Flask and a styled chat interface that classifies user intent into recipe, substitution, tips, roast, and general categories. It uses a hybrid generation strategy: a local GPT-2 model adds persona flavor while a Groq-hosted model produces the main response with context and intent-specific prompting.
Context and Post-Processing
Users can upload PDF or DOCX files, and extracted content is injected into later prompts as lightweight reference memory. Responses are post-processed to enforce recipe structure, preserve persona tone, and recover with fallback templates when the generated format is incomplete.
Export and User Experience
The app supports ingredient and step exports to PDF and DOCX, click-to-speak browser text-to-speech, and async chat interactions from a single-page UI. The result is a practical cooking assistant pipeline with prompt control, retrieval-like document context, and production-friendly output formatting.
Project Gallery
5 items




Explore More Projects
Swipe or click arrows to see other work
















