Alfonso – Python, Data Science, Machine learning

NLP-powered intelligence framework for government users, turning large-scale news data into actionable leads and evidence through news scraping, entity detection, relation extraction, and unsupervised learning with SpaCy, BERT, and REBEL.

• Led discussions with client and internal subject-matter experts to gather and refine project requirements;
• Formulated the initial product architecture design, establishing the technical foundation for the intelligence framework;
• Developed data scraping scripts to access and ingest supplementary data sources;
• Applied a combination of NLP and machine learning techniques to extract relevant entities and relationships from large-scale news data;
• Created top-level summaries to improve end-user accessibility and ease of use;
• Visualized analytical findings as a network graph to enable clearer insight into entity relationships;
• Communicated results to the client and incorporated feedback through iterative model improvements.

NLP/ML-based intelligence framework for evidence collection and investigative lead generation to support government decision-making.

• Managed a 6-member team of data scientists supporting investigative personnel across multiple departments;
• Automated the acquisition of relevant documents from multiple news sources to build a scalable ingestion pipeline;
• Trained, evaluated, and fine-tuned machine learning models to extract critical information from text-based sources;
• Deployed a Retrieval-Augmented Generation (RAG) LLM to enable flexible, user-friendly exploration of multi-document data;
• Developed a cross-document entity network analysis to uncover hidden relationships and patterns;
• Ensured analytical outputs informed senior leadership decisions and drove the initiative's expansion across multiple departments.

NLP/ML-based intelligence framework producing evidence and investigative leads to support case analysis and decision-making for government users.

• Managed a 6-member team of data scientists supporting investigative personnel across multiple departments;
• Automated the acquisition of relevant documents from multiple news sources to build a continuous data ingestion pipeline;
• Trained, evaluated, and fine-tuned machine learning models to extract critical information from text-based sources;
• Developed a cross-source network analysis leveraging entity-level relationships to uncover hidden connections and patterns;
• Deployed a Retrieval-Augmented Generation (RAG) LLM to enable flexible, user-friendly exploration of multi-document data;
• Ensured analytical findings informed senior-level decision-making and drove product expansion across multiple departments.

R-based seaborne commodity tracking platform for analyzing trade flow shifts, monitoring tanker movements, and detecting anomalous activity across multiple maritime data sources.

• Designed and implemented an R-based seaborne commodity tracking platform to analyze trade flow shifts in response to policy changes;
• Automated the consolidation of multi-source data using APIs to enable scalable and reliable ingestion pipelines;
• Performed geospatial analytics to estimate tanker movements and reconstruct shipment trajectories;
• Applied analytical methods to identify anomalies and uncover potential illicit activity in trade flows;
• Delivered insights spanning macro-level export trends to individual shipment tracking;
• Produced high-visibility reports that informed cross-departmental analysis and supported engagement with international government stakeholders.

Maritime commodity and illicit behavior tracking platform for analyzing trade flow shifts, monitoring shipments, and uncovering anomalous or illicit activity across multiple data sources.

• Designed and implemented a seaborne commodity tracking platform to analyze trade flow shifts in response to policy changes;
• Automated the consolidation of multi-source data using APIs to build scalable and reliable ingestion pipelines;
• Performed geospatial analytics to estimate tanker movements and reconstruct shipment trajectories;
• Applied analytical methods to identify anomalies and uncover potential illicit activity in trade flows;
• Delivered insights spanning macro-level export trends to individual shipment tracking;
• Produced high-visibility reports that informed cross-departmental analysis and supported engagement with international government stakeholders.

Analytics for hospital customer service and operations to improve patient experience and identify operational inefficiencies.

• Developed an NLP-based analysis of hospital survey comments, applying categorization and sentiment analysis to generate actionable insights;
• Mapped patient feedback to customer service elements, identifying key issues such as wait times and service quality;
• Established processes for continuous model evaluation and improvement, incorporating regular client feedback;
• Authored a white paper outlining methodology, findings, and future opportunities to inform next-phase development;
• Built an R Shiny dashboard to visualize key hospital metrics, including bed occupancy, readmission rates, and discharge efficiency.

Fraud, Waste, and Abuse Analytics for a government-sector client.

• Maintained and improved a production fraud detection pipeline, ensuring reliability and scalability;
• Automated ingestion and transformation of unstructured SQL data to support reporting and machine learning workflows;
• Identified fraudulent activity through pipeline models, enabling rapid case escalation to the Veterans Affairs fraud team;
• Trained and deployed machine learning models to detect payment redirect fraud, improving overall detection performance;
• Designed and executed A/B tests to evaluate process improvements, including metrics definition and statistical analysis;
• Built a fraud network dashboard in R Shiny to support connected case analysis and uncover illicit behavior patterns.

Fraud analytics for detecting, investigating, and preventing illicit activity in financial workflows

• Maintained and enhanced a production fraud detection pipeline to support ongoing monitoring and analysis;
• Automated ingestion of unstructured data from production databases via SQL, transforming it for reporting and machine learning workflows;
• Leveraged the pipeline and models to identify fraudulent activity;
• Trained, tested, and operationalized models to detect payment redirect fraud, continuously improving detection performance;
• Designed and executed A/B testing frameworks, including metric development, experiment design, and statistical evaluation;
• Developed a fraud network dashboard in R Shiny to enable connected case analysis and support deeper investigations into illicit behavior;