Hackathon Portal
AI Tinkerers - Paris
Sign in

Paris Vibe-a-thon: “Build a One-Person Unicorn” — November 15, 2025

Sign in
Engineering Lead at Stealth AI Startup - Stealth AI Startup
Own AI-native architecture and backend pipelines remotely with no equity gimmicks.
All Jobs
Final round winners have been announced. View Results
Team

CustomersAI

Project Concept

Bringing ai agent to all companies

Entry

Status: Submitted

Last saved: November 15 at 6:07 PM CET

View Entry

Team Roster

Message board not available for this team yet.

Yvan Kondjo Team Lead

Data scientist at Esigelec
Team Composition & Responsibilities Solo Developer Project - All aspects developed by one person. Technical Areas & Sponsor Tool Integration 🤖 AI/LLM Integration (Mistral AI) Implemented LangGraph autonomous agent architecture with tool orchestration Integrated Mistral AI API for primary LLM inference: mistral-small-2506 for main agent conversations and tool decisions mistral-large-latest for conversation summarization when context exceeds token limits Mistral reranking for selecting top-3 most relevant chunks from search results Built prompt engineering system with dynamic FAQ and date injection into system prompts 🔍 Embeddings & Vector Search (Google Gemini + Qdrant) Integrated Google Gemini Embeddings API (models/gemini-embedding-001): Configured with output_dimensionality=1536 for high-quality semantic embeddings Used for document chunks, website content, and search queries Implemented normalization for consistent vector representations Set up Qdrant Cloud vector database: Created knowledge_base collection with 1536-dimension vectors Implemented semantic search with user-based filtering Built RAG pipeline: query → embedding → Qdrant search → Mistral rerank → context injection 📅 Appointment Booking (Cal.com) Integrated Cal.com API for autonomous appointment scheduling: GET /availability endpoint for checking free time slots POST /bookings endpoint for creating appointments Built two LangGraph tools: check_availability_tool and create_booking_tool Implemented safety logic: agent MUST check availability before creating bookings Environment-based configuration (CAL_API_KEY, CAL_EVENT_TYPE_ID) 💾 Database & Storage (Supabase) Built entire backend data layer with Supabase: PostgreSQL database for conversations, messages, documents, FAQs, escalations Storage buckets for uploaded files (PDFs, DOCX) Real-time document status tracking (pending → processing → processed) Conversation persistence with metadata Created database schema migrations for all tables 📱 Instagram Integration (Meta Graph API) Implemented Instagram Graph API webhook integration: Webhook verification endpoint for Meta Message reception and parsing Automated response sending via Graph API Instagram account configuration UI in frontend 🎨 Frontend Development (React + shadcn/ui) Built complete React/TypeScript dashboard: Knowledge management (document upload, website crawling, FAQ editor) Playground for testing agent Settings page for AI configuration Instagram integration setup Real-time status updates with TanStack Query ⚙️ Backend Architecture (FastAPI) Designed and implemented async FastAPI backend: Document ingestion pipeline with async workers Website crawling with BeautifulSoup RESTful API for all operations Escalation system for human handoff Error handling and logging Sponsor Technologies Summary Sponsor Specific API/Product Usage Mistral AI mistral-small-2506, mistral-large-latest Primary LLM, summarization, reranking Google Gemini gemini-embedding-001 API 1536-dim embeddings for RAG Supabase PostgreSQL + Storage SDK Database, file storage, auth Qdrant Qdrant Cloud Vector search, knowledge base Cal.com Availability & Booking APIs Appointment scheduling Meta Instagram Graph API DM automation All development, integration, testing, and deployment was completed solo within the hackathon timeframe.
Jacob Kondjo is a Data Scientist Apprentice at Renault Group, based in Guyancourt, Île‑de‑France, France. With three years of experience, he applies his expertise in big data and data science to automotive challenges. Jacob earned a Diplôme d'ingénieur in Big Data and Data Science from ESIGELEC and completed his preparatory engineering studies at PrépaVogt. In addition to his apprenticeship, he works as an AI Engineer focused on application development, blending practical engineering with advanced analytics. His academic and professional background positions him to contribute to data‑driven innovation within the automotive sector.
Data Science, AGI, LLM, Robotics
SocialSyncAI — multimodal agent automating social content creation, scheduling, commenting and DMs (FastAPI/Celery backend, Next.js frontend).