Maximilien Kpizingui

• Implemented chatbot using Llama 3.2 3B foundation model and fine-tuned for specific use cases • Developed search analyst agents and business analyst agents using Composio framework • Implemented research AI agents using Camel framework for advanced automation • Fine-tuned Llama models (3.2 3B, 3.3 7B multimodal) on custom datasets using Unsloth framework • Implemented RAG systems for educational content generation and deployed using FastAPI and PostgreSQL • Integrated FastAPI-based inference APIs with JWT-based authentication and role-based access control • Designed scalable architectures to handle multiple models and concurrent user requests efficiently

Deployed no-code AI platform in production environments using dokploy, ensuring scalability, reliability, and high availability • Designed and implemented containerized workflows using Docker, orchestrating 8+ microservices to work together seamlessly • Conducted performance assessments and optimized Docker resource allocation to achieve faster deployment times and improved application performance • Implemented real-time monitoring and logging solutions using Prometheus, Grafana, and CloudWatch to ensure high visibility and reliability• Configured AWS infrastructure including Route53 for DNS management, Elastic Load Balancers for traffic distribution, and Certificate Manager for SSL/TLS • Integrated external APIs within the platform to automate data exchanges and extend functionality through custom integrations • Enhanced CI/CD pipelines using dokploy to facilitate automated testing and deployment, increasing deployment frequency • Maintained version control practices using Git, GitLab, and GitHub for efficient code management and collaboration

Built custom chatbots for startups using Llama Index, RAG, and DeepLake on corporate datasets • Fine-tuned LLMs for mask learning, causal language modeling, and sequence-to-sequence tasks • Implemented generative AI applications using LangChain with GPT-3.5 turbo and Hugging Face models • Developed machine learning models using Scikit-learn, PyTorch, and TensorFlow for various business applications • Deployed AI/ML models to production using MLOps best practices and automated pipelines • Implemented data pipelines using Apache Airflow for continuous integration and training on new datasets

Architected scalable AI agent systems using microservice architecture with dedicated Chatbot, RAG, calls and Vector Database services to unify communication in Dodo Technologies suite • Configured Docker Compose orchestration for 8+ containerized services including Nginx API gateway and Redis cache • Implemented secure service communication patterns with SSL/TLS termination and proper authentication • Optimized AI service performance through intelligent load balancing and caching mechanisms using Redis • Set up proper environment variable management and secret handling for sensitive API keys and credentials• Implemented monitoring and logging solutions for containerized applications

Developed machine learning models to predict NFT floor prices using historical data from multiple APIs • Implemented data preprocessing pipelines using pandas, numpy, and scikit-learn for model training • Used Merlion, DART, and Sktime libraries to train and evaluate time series forecasting models • Created model endpoints using FastAPI and deployed using Docker containers • Implemented NFT trading bots with automated decision-making capabilities