DriftPainel
Monitor drift between training data and current data to decide when to observe, keep or retrain models.
Includes statistical profiling, global score, feature ranking and no_action, monitor and retrain states.
LUCAS GARCIA / AI · DATA · SOFTWARE
32 audited repositories and 7 real case studies with verifiable code, decisions, metrics and limitations.
Select an atlas region
INTELLIGENCEModels, RAG, baselines and evaluation.
Each system record starts with the problem and ends in public links, stack, evidence and limitations verified in the repository.
Monitor drift between training data and current data to decide when to observe, keep or retrain models.
Includes statistical profiling, global score, feature ranking and no_action, monitor and retrain states.
Analyze CSV files repeatably, reveal quality issues and generate practical cleaning recommendations.
Detects nulls, duplicates, cardinality issues, inconsistent types, empty and constant columns, problematic names and outliers.
Organize parish administration, pastoral workflows, public communication and governance with secure module-based access.
Implemented vertical flow: create person, view person, edit person and audit changes.
Answer operational incidents from local runbooks with diagnosis, checklist and traceable sources.
Delivers diagnosis, resolution steps, copyable checklist, sources, confidence, feedback and operational metrics.
Collect, normalize and analyze international remote jobs from public sources and structured files.
Includes Bronze/Silver/Gold, Data Quality, Schema Drift, Data Lineage, CSV/Parquet/Markdown/HTML exports and a safe scraping policy.
Estimate monthly income from socioeconomic and occupational features with methodological rigor and interpretability.
Documented evaluation on 25,801 observations: CatBoost reached 0.5913 log-scale R² and the hybrid model reached 0.4906 R² on the original scale.
Turn a manual full-stack deployment into declarative, repeatable infrastructure on AWS.
The repository contains 96 files, three Terraform modules and Docker images for frontend and backend, plus the full-stack application.
Select an atlas region
INTELLIGENCEModels, RAG, baselines and evaluation.
The focus is making evaluation easier for recruiters: open code, READMEs, pipelines, tests and objective case studies.
Experience ↗Ifes · Campus Colatina · PRPPG/PICTI
LEDS · Ifes
Ifes · Campus Colatina
A record of scientific-initiation work combining information retrieval, LLMs and experimental evaluation in the intellectual-property domain.
How global score, feature ranking and history help decide between monitoring and retraining.
Scores, recommendations and reports reduce the distance between technical diagnosis and action.
Controlled collection, manual review and lineage protect pipeline quality.
Decisions get metrics, logs, tests and limits.
Complexity enters when it proves value.
Data, APIs, interfaces and documentation share quality.