Miguel Carmona

Advanced Modeling & Bayesian Optimization:

• Led the transition from frequentist MMM to Bayesian regression using MCMC & Monte Carlo simulations.
• Developed machine learning models to enhance advertising targeting and campaign effectiveness.
• Boosted model accuracy (R² > 90%) and reduced modeling time by 70%.
• Built internal libraries and global simulation tools adopted by international teams.
• Implemented and optimized algorithms for real-time data processing and predictive analytics.
• Ensured model accuracy through rigorous testing and validation processes.

Generative AI Development:

• Designed and deployed multiple AI agents in production:
• Mike: RAG-based assistant for corporate knowledge, used internally and sold to clients.
• Finance Agent: autonomous chatbot querying internal DBs via email integration.
• CCS Agent: Chainlit-based conversational interface with data querying and visualization.
• Deployed APIs with Flask and frontend interfaces with Chainlit and React.
• Ran A/B tests with randomized segmentation for model validation.

Infrastructure & Deployment:

• Built and deployed solutions on GCP (Cloud Run, Kubernetes, BigQuery, Firebase, Artifact Registry)
• Used CI/CD, Docker, and Airflow to automate and scale pipelines.
• Contributed to the design of scalable architecture for large-scale machine learning applications.

Cross-functional Leadership & Communication:

• Acted as PM, coordinating a team of 3 developers.
• Mentored junior engineers on best practices in machine learning development and deployment.
• Aligned delivery with product, design, and dev teams across multiple initiatives.
• Collaborated with cross-functional teams to gather requirements for data-driven projects.
• Presented MMM results and campaign insights to key clients (Vodafone, Lowi and more).
• 2nd place – European internal hackathon (AI-focused).

• Optimized real estate pricing predictions: Implemented regression models to predict property prices across Spain, improving prediction accuracy by [87 - 93%].
• Designed the Spanish Real Estate Market Index (IMIE): Built a proprietary index to track market trends and forecast housing price behavior nationwide, providing actionable insights for stakeholders. Deployed via dynamic web interface.
• Maintained property appraisal systems in Portugal: Designed and maintained a centralized database responsible for processing the vast majority of property valuations in the country, ensuring reliability and operational efficiency.