Pablo Baretto

Senior Software Engineer with over 20 years of experience as a Solutions Architect and Data Engineer. In recent years, I have specialized in developing AI applications and consulting companies on how to build reliable, production-grade AI solutions that actually reach production.

I help companies to plan practical AI initiatives, design production-grade architectures, evaluate AI systems with real data and accelerate their AI engineering efforts.

Request a Consultation

Experience

2021 – Present: AI Architecture and Development Consultant (Freelance)

I engage in long-term consulting commitments (up to 2 years) with mid-market companies and venture-backed startups, designing system architectures and leading development of core AI-driven functionalities. Projects focus on building production-grade, high-accuracy AI systems in regulated or data-intensive domains.

Healthcare Startup (USA)
AI-powered compliance platform that analyzes mental-health provider reports to ensure insurance compliance.

  • Problem: Insurance payers reject claims at high rates due to imprecise or non-compliant language, causing revenue loss and administrative burden.
  • Solution: Architected and developed the core generative AI compliance engine. Implemented a data-driven AI evaluation framework + human-in-the-loop error analysis to overcome LLM hallucinations and biases.
  • Results: Achieved sustained 98% accuracy. Only ~2 reports per day now require human review, drastically reducing manual effort while minimizing claim denial risks.

Legal SaaS Startup (Spain)
Agentic RAG-based AI chatbot SaaS that answers legal questions for users in Ecuador, grounded in current national laws (owned by a prominent law firm).

  • Problem: Manual legal research was time-intensive and error-prone; keeping the knowledge base updated was slow and risked hallucinations or SLA penalties.
  • Solution: Built an automated real-time data processing pipeline for laws + developed an agentic RAG application with precise citations and rigorous AI evaluation frameworks.
  • Results: Reached 99% response accuracy, reduced human review to 1% of cases, and cut time spent answering client questions by ~90%.

Matchbook AI (USA)
SaaS master data management platform integrating and unifying company/asset data for S&P 500 clients.

  • Problem: Fragmented, inconsistent records across sources led to poor insights, duplicates, and SLA breaches. Legacy Elasticsearch matching required heavy manual intervention.
  • Solution: Designed and led development of a new ML/AI-based matching engine + full automation of the legacy Elasticsearch system.
  • Results: Handles ~500 million initial records + ~1 million daily updates (15% above targets) and powers the core platform for ~50 S&P 500 clients. Legacy automation eliminated all manual processes, cut processing time by 35%, and reduced SLA breaches to zero.

Sonarplatform.io (EU)
SaaS platform for blockchain transaction analytics and AI-driven trading insights.

  • Problem: Indexing billions of transactions with low-latency queries was technically challenging and resource-intensive.
  • Solution: Architected and led development of the core blockchain data platform (scalable ingestion, indexing, and query layers).
  • Results: Indexes ~1 billion transaction records across 10+ Ethereum networks, handles ~3,000 tx/s in real time, and delivers API responses in <800 ms (p95).

AI driven Startup (USA)
SaaS platform for personalized AI chatbots/companions using users’ own data.

  • Problem: Low-latency, high-accuracy search over large personal data volumes was difficult with traditional indexing.
  • Solution: Architected the personal information search subsystem (high-throughput indexing pipeline + hybrid lexical + vector search backend).
  • Results: Processes up to 200 MB/s for thousands of users; search delivers highly accurate results in <500 ms.

2015 – 2021: Logicalis (Spain)

IT services and consulting firm (part of the global Logicalis Group) focused on AI, data analytics, cloud, and digital transformation. Progressed from Cloud Solutions Architect to Data Analytics Solutions Architect.

Key Projects (Telecom & Enterprise Clients): - Orange: Architected and led development of a unified log and metrics processing platform (Elasticsearch-inspired stack).
Results: Reduced MTTD by ~55%, MTTR by 45%, unplanned downtime by 30–40%, and manual log analysis effort by >60%. - Delivered multiple production implementations and PoCs for Vodafone, Telefónica, FNMT, Grupo Piñero, La Caixa, Ocaso, AENA, and Catalana Occidente, enabling centralized observability and faster digital operations.

2006 – 2015: Senior Data Analytics Consultant

Focused on analytics for IT monitoring and support systems.

  • FacilDesk (Founder): Open-source IT monitoring (Elasticsearch) + ticketing (OTRS) with analytics for Latin American mid-sized firms.
    Notable clients: Mitsubishi Motor Company (+67% customer satisfaction), Agroamerica, Cerámicas Caribe, Bancaribe (43% faster incident response).
  • Sun Microsystems Contractor (Venezuela): Data-centric monitoring for telcos (CANTV, Movilnet, TelefĂłnica) and oil & gas (PDVSA, ENI).

2001 – 2006: Sun Microsystems (Venezuela)

Java trainer and consulting in northern Latin America.

Certifications

  • Google Cloud Professional Data Engineer (2020)
  • Elasticsearch Certified Engineer (2020)

Education

  • Systems Engineer, Universidad de Los Andes (Venezuela), 2001
  • Master’s in Marketing Management, Universidad Complutense de Madrid (Spain)

Request a Consultation