Have you built autonomous AI agents for code generation, refactoring, or code analysis?

Yes. Since 2024 I design, build and operate a fleet of autonomous agents that maintain a 1,300+ repository portfolio: a dual-agent documentation system (evaluator + executor) that audits repos and autonomously writes and refactors documentation, and a machine-enforced adversarial review coordinator with architect, critic and verifier agents on distinct LLM families and a hash-chained audit ledger. In its latest production run the critic layer caught 7 defects in a deploy plan before it reached production.

How much experience do you have with Python and RAG system design?

Python: 15+ years. RAG: 3+ years in production, including Voyage embeddings with Qdrant (HNSW) delivering sub-200ms semantic search, and an enterprise RAG platform over a 100K+ document corpus with LangChain and Pinecone.

Which AI tools do you use in your day-to-day work?

Claude Code and Claude Cowork as primary engineering environment; custom MCP servers I built and operate; LangGraph multi-agent orchestration with model-family separation; Composio tool routing across GitHub, Vercel and Cloudflare; every production change gated by agent-driven triple validation.

Can you describe an agentic system you personally built and operated in production?

An adversarial multi-agent deployment gate: an architect agent produces a change plan, two critic agents on different model families attack it demanding command-level evidence, and a verifier independently re-validates every claim. State transitions are hash-chained and deploy hooks execute only on a cryptographically verified approval. I am the sole architect and operator; in its most recent production run it rejected the initial deploy plan with 7 evidenced defects, fixed before production.

What is your level of English?

Fluent (C1), with daily client-facing delivery for EU and US Fortune 500 accounts.

THIAGO ANTAS

Principal AI Solutions Architect | Agentic AI (LangGraph, MCP) · Production RAG · LLMOps | AWS Bedrock, Azure AI Foundry & Multi-Cloud

Lisbon, Portugal · thiago@thiago.pt · Remote B2B worldwide · EU/US time-zone coverage · B2B via Thiago Antas Unip LDA
linkedin.com/in/tfantas · github.com/tfantas · hackerrank.com/profile/tfantas · www.tfantas.io · ORCID 0009-0000-9408-5663

Professional Summary

I design and ship agentic AI systems that run in production — not demos. Principal-level Solutions Architect with 21+ years across insurance, financial services, energy, and public sector, including Fortune 500 client delivery at Capgemini. Over the last 3 years, my focus has been end-to-end GenAI delivery: multi-agent systems with LangGraph, MCP, and A2A; production RAG; LLMOps and evaluation harnesses; and AWS Bedrock and Azure AI Foundry on multi-cloud Kubernetes (EKS, AKS). I have delivered cloud-native platforms processing 2M+ events/day and cloud migrations that reduced infrastructure costs by 45%. I also co-founded a startup that scaled to 50,000+ active users. Available for remote B2B consulting and Staff/Principal AI Platform roles worldwide, with EU/US time-zone coverage. On the Microsoft stack specifically, I have delivered production conversational AI on Azure AI Foundry with Azure AI Search as the retrieval and grounding layer — including a Portuguese public-sector assistant and a telecom enterprise agent — industrialized through Azure DevOps CI/CD and signed off on latency, answer quality, and cost before launch. I work end to end across pre-sales, POCs, pilots, and production delivery, translating business requirements into Azure-native solution architectures.

Core Competencies

Agentic AI & Multi-Agent Systems: LangGraph, MCP, A2A, agent orchestration, tool calling, guardrails, evaluation, and observability
Production RAG & LLMOps (3+ years in production): Vector and hybrid retrieval, context engineering, evaluation harnesses, prompt engineering, cost/latency optimization, and AI FinOps
Cloud AI Platforms: AWS Bedrock, Azure AI Foundry & Azure OpenAI, GCP Vertex AI, AWS EKS, AKS, Kubernetes, Terraform, and multi-cloud platform engineering
Microsoft Azure AI Stack: Azure AI Foundry, Azure AI Search (vector, semantic, and hybrid retrieval with grounding), Azure OpenAI, Azure Functions, Cosmos DB, Entra ID, Bicep, and Azure DevOps CI/CD for enterprise-grade, compliance-aware AI delivery
Pre-Sales & End-to-End Delivery: Solution discovery and architecture, POCs and pilots taken through to production, decomposition into ingestion, retrieval, orchestration, and guardrail layers, and launch readiness measured on latency, answer quality, and cost
Executive Advisory: AI roadmaps, EU AI Act alignment, build-vs-buy decisions, platform operating models, and enterprise governance
Apple Platform & Native Development: Swift 6.2, SwiftUI, Liquid Glass, AppKit, On-Device LLM/STT, Real-Time Transcription & Translation (iOS 26 / macOS)
Architecture & Development: Event-Driven Architecture, Microservices, CQRS, DDD, API Design (REST, GraphQL, gRPC)
AI-Native Daily Workflow: Claude Code / Claude Cowork as primary engineering environment; custom MCP servers built and operated (docs-mirror agent exposed via MCP stdio + A2A HTTP); multi-agent orchestration with model-family separation (Opus/Sonnet/Haiku critics); Composio tool routing across GitHub, Vercel, and Cloudflare; every production change gated by agent-driven triple validation (API + checksum + visual)
Languages & Frameworks: Python (15+ years), Rust, TypeScript, Java, Go | LangChain, FastAPI, Spring Boot, Node.js, React

Professional Experience

Capgemini — Lisbon, Portugal

Senior AI & Software Architect / Tech Lead | 2018 – Present (7+ years)

Designed multi-agent architectures with LangGraph, MCP, and A2A, including guardrails, evaluation, and observability for enterprise production environments
Delivered production RAG pipelines with vector and hybrid retrieval, context engineering, and cost/latency optimization on AWS Bedrock and multi-cloud Kubernetes
Implemented LLMOps foundations and evaluation harnesses to support safer, measurable GenAI delivery
Built cloud-native, event-driven platforms processing 2M+ events/day
Led cloud migration and platform modernization initiatives that reduced infrastructure costs by 45%
Advised senior stakeholders on AI roadmaps, EU AI Act alignment, build-vs-buy decisions, and enterprise governance
Applied deep enterprise architecture, distributed systems, Java/Spring, and Python experience across regulated public-sector, insurance, financial-services, and energy programs
Led a technical team of 16 engineers across enterprise delivery programs
Mentored engineering teams in cloud architecture, platform engineering, DevOps, and production AI delivery
Designed and delivered a Microsoft-based conversational assistant for a Portuguese public-sector client on Azure AI Foundry with Azure AI Search, focused on secure enterprise retrieval, controlled access, and compliance
Built a telecom enterprise conversational agent on Azure AI Foundry with Azure AI Search, combining retrieval-augmented generation, agent orchestration, and guardrails so the agent returns grounded, enterprise-safe responses from only the most relevant sources
Mapped business requirements to Azure-native solution architectures and industrialized delivery through Azure DevOps pipelines covering CI/CD, automated testing, and deployment
Owned engagements end to end across pre-sales, POCs, pilots, and production delivery, validating RAG and agent flows and gating launch on latency, answer quality, and cost

Above-Net — Rio de Janeiro, Brazil

Solutions Architect | 2018 (1 year)

Solution architecture for Java SE and AWS-based enterprise systems (EC2, ELB, RDS)

Druid — Rio de Janeiro, Brazil

Java Architect / Solutions Architect | 2017 – 2018 (1 year)

Designed Netflix OSS microservices architecture (Eureka, Hystrix, Config Server, API Gateway) on AWS, with Docker and Jenkins CI/CD

Urban Mobility Startup — Rio de Janeiro, Brazil

Co-Founder & CTO | 2016 (1 year)

Scaled platform from MVP to 50,000+ active users within 18 months, securing R$100K seed funding
Built real-time geolocation system with ML-powered route optimization, reducing delivery times by 30%
Led technical team of 8 engineers; designed full-stack architecture (React Native, Node.js, PostgreSQL, Redis)

Noorden Group — Rio de Janeiro, Brazil

Java Architect & Agile/DevOps Coach | 2015 – 2017 (2 years)

Led SCRUM adoption and CI/CD standardization (Maven, Jenkins) on-site at SulAmérica Seguros, with TDD coaching using JUnit, XMLUnit, and Selenium

Petrobras — Rio de Janeiro, Brazil

Senior Software Engineer | 2012 – 2015 (3 years)

Delivered a corporate land-control system at TIC-ENG (Engineering IT & Telecommunications)

Gazeus Games

Senior Software Engineer | 2011 – 2012

Senior Java engineering for online gaming platforms

ProBid Consultoria e Serviços

Senior Developer Analyst | 2010 – 2011 (1 year)

Senior Java development and systems analysis across consulting engagements

Politec Global IT Services

Senior Software Engineer | 2008 – 2010 (2 years)

Senior Java engineering on large-scale IT outsourcing programs

Tools Software / TOTVS Financial Services

Systems Analyst | 2004 – 2008 (5 years)

Systems analysis and Java / PL-SQL development for financial services products

Key Projects & Achievements

Autonomous Code & Docs Agent Fleet (personal production infra, 2024–present): Designed, built, and operate a fleet of autonomous agents maintaining a 1,300+ repo portfolio: a dual-agent documentation system (evaluator + executor) that audits repos and autonomously writes and refactors documentation, and a machine-enforced adversarial review coordinator — architect/critic/verifier agents on distinct LLM families with a hash-chained audit ledger, where deploy hooks fire only on a cryptographically verified approved state. In its latest production run, the critic layer caught 7 defects in a deploy plan before it reached production. Sole architect and operator
Azure Public-Sector Conversational Assistant (Portugal): Microsoft-stack enterprise assistant built on Azure AI Foundry with Azure AI Search for retrieval and grounding — secure knowledge access with controlled permissions, compliance alignment, and a hardened cloud architecture for scalable deployment
Telecom Enterprise Conversational Agent (Azure AI Foundry + Azure AI Search): Production RAG agent with orchestration and guardrails that retrieves only the most relevant enterprise sources, delivered through Azure DevOps CI/CD and evaluated on latency, answer quality, and cost before launch
Second Brain — iOS 26 Native App: Real-time cognitive-augmentation app for Apple platforms — Swift 6.2, SwiftUI, Liquid Glass — featuring on-device speech-to-text, LLM reasoning, and live translation through a multi-agent architecture
Brain Core Engine: High-performance retrieval core powering Second Brain — Voyage 3.5 embeddings with a Qdrant vector database and HNSW indexing, delivering sub-200ms semantic search
macOS Native Development Blueprint: Hardware-aware, AI/LLM-optimized blueprint for native macOS development, tuned for Intel MacBook Pro 16,4 (Late 2019 — Core i9, Radeon Pro 5600M)
Enterprise RAG Platform: Production-grade retrieval system using LangChain, Pinecone, and OpenAI embeddings for a 100K+ document corpus with semantic search
AI Agent Orchestration Framework: Custom LangGraph framework for autonomous multi-step task execution with human-in-the-loop governance
Cloud Migration & Platform Modernization: Led Kubernetes-based migration and modernization initiatives that reduced infrastructure costs by 45%
1,300+ Private GitHub Repositories: Continuous innovation portfolio across AI/ML, Agentic AI, DevOps, and enterprise architecture

Education & Certifications

Professional Certificate in AI and Machine Learning — Purdue University | Jul 2025 – Feb 2026 Purdue PCP AIML Cohort 82 — Applied Data Science, ML, Deep Learning, GenAI, RAG, Prompt Engineering
Anthropic Academy — Engineering & Platform (15 verified certificates, 2025–2026): Claude Code in Action · Claude Code 101 · Claude Platform 101 · Claude 101 · Building with the Claude API · Introduction to Model Context Protocol · Model Context Protocol: Advanced Topics · Introduction to Agent Skills · Introduction to Subagents · Introduction to Claude Cowork · Claude with Amazon Bedrock · Claude on Google Cloud · AI Capabilities and Limitations · AI Fluency: Framework & Foundations (scored 9/10) · AI Fluency for Builders
Anthropic Academy — AI Enablement & Train-the-Trainer (6 verified certificates + 1 completion, 2025–2026): Teaching AI Fluency · AI Fluency for pK-12 Educators · AI Fluency for pK-12 Train the Trainer · AI Fluency for educators · AI Fluency for students · AI Fluency for Small Businesses · AI Fluency for nonprofits — qualified to run AI-adoption enablement and workshop programmes for non-technical stakeholders
MBA in Software Engineering — UFRJ, Federal University of Rio de Janeiro | 2013 – 2014
Bachelor’s in Information Systems | 2007 – 2011 · Technical Degree in Computer Science | 2003 – 2005
Core Expertise Areas: AWS Solutions Architecture | Kubernetes Administration | Agile (Scrum Master, Scrum Developer) | DevOps & CI/CD
AI/ML (Purdue 2025–2026): Deep Learning | ML with Python | TensorFlow | GANs | Supervised/Unsupervised ML | Prompt Engineering | RAG | Databricks | ChatGPT Advanced

Technical Stack

AI/LLM: GPT-4, Claude, Llama, Mistral, DeepSeek | LangChain, LangGraph, CrewAI, AutoGen | RAG, Vector DBs (Pinecone, Weaviate, ChromaDB, Qdrant) | Voyage embeddings, HNSW | LoRA/QLoRA Fine-tuning | LangSmith, W&B
Cloud: AWS (Bedrock, SageMaker, EKS, Lambda) | Azure (AI Foundry, OpenAI Service, AKS, Functions, Cosmos DB, AI Search, Entra ID, Azure DevOps, Bicep) | GCP Vertex AI | Cloudflare Workers | Vercel, Supabase
Infrastructure: Kubernetes, Docker, Terraform, Helm, ArgoCD | GitHub Actions, GitLab CI/CD | Kafka, RabbitMQ, Redis
Apple Platforms: Swift 6.2, SwiftUI, Liquid Glass, AppKit | On-Device LLM/STT, Real-Time Transcription & Translation (iOS 26 / macOS)
Languages: Python, Rust, TypeScript, Java, Go, SQL | FastAPI, Spring Boot, Node.js, React, Next.js

Languages

Portuguese (Native) · English (Fluent — daily EU/US client-facing delivery) · Spanish

Online Profiles & Competitive Programming

GitHub — github.com/tfantas 1,300+ private repositories (AI/ML, Agentic AI, DevOps, enterprise architecture) · GitHub Developer Program Member · Achievements: Pull Shark ×3, Pair Extraordinaire, Quickdraw, YOLO
HackerRank — hackerrank.com/profile/tfantas Badges: Problem Solving 6★ · Java 5★ · Python 5★ · SQL 5★ · 30 Days of Code 5★ · React 2★
14 verified certifications — Problem Solving (Intermediate); SQL (Intermediate, Advanced); Python; Java; JavaScript (Intermediate); React; Node; REST API (Intermediate); CSS; Software Engineer (Role)
Professional Certifications (2025): Purdue PCP AIML (Cohort 82) · 21× Anthropic Academy (Claude Code, MCP, Agent Skills, Subagents, Bedrock, Vertex AI, AI Fluency) · 16× Simplilearn SkillUp — Generative AI, Deep Learning, ML with Python/R, TensorFlow, GANs, Prompt Engineering, RAG, DeepSeek, Mistral AI, ChatGPT Advanced, Data Analytics · 3× Databricks — SQL Analytics & BI, Generative AI, for Business Leaders
ORCID — 0009-0000-9408-5663

Last updated: 13 August 2026