3. Data, Science & AI

Enables creation, editing, and analysis of spreadsheets including formulas, formatting, and data visualization for data-driven tasks.

Loads relevant OpenContext documentation for AI tasks without index building, enabling context-aware responses.

Validates ClickHouse SQL syntax and tests queries without a running server, enabling efficient database query development and debugging.

Extracts complete technical property descriptions from Argentine notarial deeds, preserving boundaries, measurements, and areas across page breaks and multiple properties.

Builds production RAG systems with semantic chunking, incremental indexing, and context-aware retrieval using vector search.

Manages local HuggingFace models and datasets to avoid redundant downloads during data loading.

Designs memory architectures for AI agents to persist state across sessions, maintain entity consistency, and reason over structured knowledge.

Classifies uploaded documents (PDFs/images) into notarial categories using AI, routing to specialized extraction skills.

Expert agent for populating, cleaning, and validating a scientific exercise database using biomechanics and physiology expertise.

Extracts, structures, and validates identity data from physical documents (e.g., Argentine ID, passports) using multimodal computer vision and MRZ analysis for security verification.

Enables real-time streaming UI patterns for AI chat applications with response lifecycle handling and progress indicators.

Extracts mortgage financial conditions including loan amounts, interest rates, UVA index, and amortization details from documents.

Automates multi-perspective idea evaluation by deploying specialized AI agents from a predefined folder for expert analysis.

Provides structured pruning for Mixture of Experts (MoE) models using micro-expert energy estimation to enhance model efficiency and reduce computational overhead.

Multi-agent parallel planning framework enabling simultaneous multi-perspective design and consensus-driven implementation planning.

High-performance Rust web crawler for extracting and structuring web content, optimized for LLM/RAG pipelines with Markdown export and stealth capabilities.

Specialized entity extractor for Argentine notarial documents, handling real-world edge cases and system errors.

Automates conversion of narrative legal documents into structured RPI/Catastro-compliant metadata (XML/JSON/PDF), generating official forms and code mappings.

Enables audit log search, analytics, and dashboard building for INVOOPAY using Elasticsearch data streams and event indexing.

Specializes in database management for PostgreSQL, MongoDB, and Redis, implementing advanced data patterns for modern application development.

Optimizes SQL queries for improved database performance and efficiency in data processing workflows.

Manages Airtable records via API for CRUD operations, batch processing, pagination, and linked records in the MI Platform base.

Expert PostgreSQL database management for INVOOPAY: design schemas, write complex queries, optimize performance, and handle migrations and debugging.

Formats dates, amounts, and names per Argentine notarial protocol, converting numbers to words and standardizing DNI and surnames for public deeds.