Engineering
Sovereign AI & Edge Systems
Research and development of RAG architectures, IoT mesh networks, and distributed predictive models.
Research & Publications
Peer-reviewed work and preprints on RAG architectures, edge computing, and distributed systems.
Sovereign RAG: Architectures for Domain-Specific Retrieval-Augmented Generation
Novel approach to RAG systems with emphasis on data sovereignty and domain isolation for legal and geotechnical applications.
Distributed Predictive Models for Geotechnical Risk Assessment
Edge-deployed ML models for real-time slope stability prediction using distributed sensor networks in mining operations.
LegalTech RAG: Multi-Tenant Architecture for Chilean Legal Corpus
Scalable retrieval system design for legal document processing with tenant isolation and compliance-aware filtering.
Mesh IoT Networks for Remote Geotechnical Monitoring
Low-power mesh network architecture for sensor deployment in extreme environments with offline-first data sync.
Kana: Local LLM Framework for Domain-Specific Tasks
Lightweight framework for running quantized language models on edge hardware with domain adaptation capabilities.
Applied Architectures
Production systems demonstrating full-stack capability from hardware to cloud infrastructure.
MinerScan IoT
Personal geotechnical monitoring system with edge-deployed predictive models. Real-time slope stability analysis using distributed sensor arrays.
LegalFast Ecosystem
Scalable LegalTech architecture evolved from LegalBot Chile. Multi-tenant RAG system for Chilean legal corpus with compliance-aware retrieval.
Gilgamesh CLI
Agent orchestration hub with local memory persistence. MCP-based tool execution and cross-session context management.
Compute Partnership
We are scaling our predictive models and digital twin simulations. Actively exploring computing partnerships and infrastructure grants to process large-scale geotechnical and legal data.
Ready to Collaborate?
Open to research partnerships, compute grants, and technical advisory roles in AI infrastructure.