Lucas Sinibaldi Bonere

Recommendation & Scalability Architecture: Design of a "home-made" vector search engine (FastAPI + Qdrant) enabling similarity search in < 50ms on millions of items, increasing the click-through rate (CTR) on recommendations by 25%.

MLOps Lifecycle (Airflow & MLflow): Complete automation of model retraining and performance monitoring, reducing the "Time-to-Market" for new algorithmic iterations from 2 weeks to 2 days.

Specialized NLP & Sentiment Analysis: Fine-tuning of LLMs achieving 94% accuracy on complex opinion extraction (irony, cultural nuances), improving the relevance of overall scores per work by 15%.

Hybrid Personalization: Development of engines combining collaborative signals and content embeddings, generating a 20% increase in retention rate (Daily Active Users) through hyper-personalized suggestions.

Infra Cost Optimization: Migration to an optimized asynchronous and vector architecture, reducing the necessary compute resources by a factor of 3 compared to old collaborative filtering SQL methods.

• Development of an automated ML pipeline with Vertex AI for predictive SEO performance analysis: design and deployment of machine learning algorithms (Random Forest, XGBoost) on Vertex AI to predict organic traffic evolution and identify keyword opportunities, reducing client analysis time from 1 day to 30 minutes.

• Design of an intelligent ETL system with NLP classification enabling automatic detection of search intent on heterogeneous SEO data (Search Console, crawl logs, backlinks). Use of K-Means semantic clustering technique on query embeddings to automate query segmentation based on semantic similarity, eliminating 95% of manual errors and saving over half a day per analysis.

• Analysis and monitoring of models in production with drift tracking via RMSE, MAE, R² and MAPE metrics; implementation of an advanced monitoring system to detect early degradation of SEO model performance using these KPIs. Integration of an Isolation Forest model for anomaly detection. This system has increased the reliability of models in real-world environments and ensured the maintenance of a high performance level upon arrival of new data, thus ensuring optimal steering of SEO strategies.

Architecture & Cloud Solutions: Design, deployment, and migration of business management and accounting applications on Cloud infrastructures, ensuring high availability and service scalability.

Data Engineering: Modeling, optimization, and administration of complex relational and non-relational databases dedicated to fine-grained customer portfolio management (CRM).

Cross-Platform & Mobile Development: Strategic porting of desktop (PC) software solutions to native mobile applications using Flutter and Android Studio, improving accessibility and field user experience.

Web & UI/UX Design: Creation of modern, responsive, and user-centric web interfaces for business intelligence and data analysis dashboards.

Cross-Functional Collaboration: Translation of accounting and business needs into technical specifications, reducing friction between product and technical deployment.