We are looking for an experienced AI developer to build an AI-powered smoking cessation assistant designed for healthcare professionals (pharmacists and physicians). The assistant will help healthcare professionals guide patients who want to quit smoking by analyzing their profile and smoking habits, then providing evidence-based suggestions. A critical requirement of this system is that the AI must only generate answers based on predefined scientific sources (PDF documents and similar resources). If the user asks a question that is not covered in the provided sources, the AI must not generate an answer and instead respond with a message such as: “This information is not available in the current sources.” This system will essentially function as a Retrieval Augmented Generation (RAG) system. Core Features 1. Source-Based AI (RAG System) The AI must: - Retrieve answers only from uploaded documents (PDF, medical resources, etc.) - Avoid hallucinations - Refuse to answer questions if the information is not found in the sources - Display the source reference used for the answer The interface should include a panel showing the available sources used for responses. 2. Patient Assessment (Pre-Consultation Questions) Before the assistant provides recommendations, it should ask several assessment questions to evaluate the patient's smoking habits and personal information. Examples: -Basic Information - Age - Gender - Occupation - Health conditions - Current medical issues Smoking Dependency Evaluation (Fagerström-like test) Example questions: - How many cigarettes do you smoke per day? - How soon after waking up do you smoke your first cigarette? - Do you find it difficult to refrain from smoking? - Which cigarette would be the hardest to give up? The system should calculate a smoking dependency score based on the answers. 3. Personalized Recommendations Based on the collected data, the assistant should provide personalized guidance for smoking cessation. Recommendations should vary depending on: - age - gender - smoking dependency level - health condition - financial situation 4. Product Recommendation At the end of the interaction, the assistant should suggest appropriate smoking cessation products. Examples: - Students → more affordable options - High dependency → stronger treatment options - Health issues → medically suitable alternatives The recommendation should also include motivational messaging. 5. Motivational Messaging System The assistant should include motivational messages tailored to user characteristics. Examples: - For female patients: -- improved skin health -- improved fertility -- better long-term wellness - For male patients: -- improved physical performance -- higher energy levels -- cardiovascular benefits This motivational framework should be expandable over time. 6. Learning and Improvement System The assistant should collect interaction data in order to improve over time. Examples of data to capture: - user objections - frequently asked questions - situations where responses were not satisfactory This data should be stored in a structured way so the system can be continuously improved. Important: the system should not automatically change medical knowledge, but the conversation data should be available for future training or optimization. 7. Target Users This system will not be publicly available. It will only be used by: - pharmacists - physicians The system will therefore function as a clinical consultation support tool. Interface Requirements - The interface should be simple and clean. - Basic layout: -- Left side: Chat interface -- Right side: Source documents panel - References used for answers Technical Requirements - The developer should have experience with: - Large Language Models - RAG systems - Vector databases - Prompt engineering - PDF processing - Chat interface development - API integration Preferred technologies (open to suggestions): - LangChain or LlamaIndex - OpenAI or similar LLM providers - Pinecone / Weaviate / Chroma - React or Next.js Deliverables The project should include: - Working AI assistant - RAG-based document retrieval system - Patient assessment question flow - Smoking dependency scoring system - Personalized recommendation engine - Product recommendation logic - Source citation system - Interaction data logging - Basic web interface Additional Notes - The AI must only answer based on the uploaded sources - No hallucinated medical content - Medical accuracy is critical - Developer is welcome to propose the best architecture for this system Proposal Requirements When submitting your proposal, please answer the following questions: 1- How long would it take you to complete this project? 2- What is your estimated budget for this project?