Personal details

Matheus S. - Remote full-stack developer

Matheus S.

Senior Full-Stack & AI Engineer
Based in: 🇧🇷 Brazil
Timezone: Brasilia (UTC-3)

About

Hi there! I'm Matheus, a Senior Full-Stack Software Engineer and currently CTO at Divimah, with nearly 10 years building software and 12 years in tech. I'm strongest in Angular, TypeScript and Node.js/NestJS, with substantial hands-on work in React, Next.js and React Native (Expo), and I've taken products from architecture to production for travel, fintech, healthcare, banking and telecom clients.

I focus on building maintainable, scalable software with strong frontend and backend foundations: responsive UIs, secure integrations, clear data flows, and measurable business outcomes. I've shipped real-time collaboration and workflow systems using event-driven patterns, cross-platform mobile apps, and the reliability work that keeps them running: caching, distributed locking, CI/CD hardening, incident response and safe data recovery on live databases.

AI-native engineering is how I work every day. Since 2025, Claude Code and Codex are core tools in my workflow, and I build the system around them, not just the prompts:

  • Agent skills and MCP: 60+ reusable agent skills (domain playbooks, guarded database operations with dry-run first, code review, ticket automation, iOS release flows) plus MCP integrations wired into the agents' toolset
  • Multi-agent delivery: spec-driven builds where parallel agent teams work in isolated git worktrees, each task routed to the right model tier, with an adversarial review gate before any PR
  • Self-verifying agents: headless test rigs and scripted end-to-end app journeys with screenshots, so every change is proven by exit codes and evidence, not by an agent's word
  • RAG and LLM features: hybrid retrieval (BM25 + embeddings, rank fusion, reranking, golden-set evals), LLM-assisted suggestions and classification flows, and the AI roadmap at Divimah (ML recommendations in production, LLM-assisted operations)

My technical backbone:

  • Frontend: Angular (Signals, NgRx, RxJS), React, Next.js, TypeScript, Ionic/Capacitor, React Native (Expo), Tailwind CSS
  • Backend: Node.js, NestJS, Python, Java, .NET, REST APIs, WebSockets (Socket.IO), event-driven architecture
  • Data: MongoDB, PostgreSQL (pgvector, row-level security), SQL Server, Oracle, Redis, Prisma, TypeORM
  • AI: Claude Code, Codex, OpenAI and Anthropic APIs, LLM agents, MCP, RAG, embeddings, prompt engineering
  • Cloud / Ops: AWS, DigitalOcean, Vercel, Railway, Cloudflare (Workers, R2), Docker, GitHub Actions, CI/CD, iOS TestFlight releases
  • Execution: technical leadership, roadmap and sprint planning, code reviews, mentoring, clear communication in English, Portuguese and Spanish

I thrive in fast-moving teams where delivery quality matters: translating product goals into shipped software, reducing ambiguity, and balancing speed with maintainability to create products that truly move stakeholders and users.

Work Experience

CTO
Divimah | Feb 2025 - Present
Python
Node.js
MongoDB
Angular
API
PostgreSQL
Unit Testing
WebSocket
Redis
TypeScript
Jira
Docker
React
Jest
DigitalOcean
RxJS
NgRx
Responsive Web Development
API Integration
Engineering Management
RESTful API
NestJS
Prisma
Rest api design
GitHub Actions
Playwright
Ionic capacitor
LLM
RAG
Sprint management
React (next.js)
Claude code

Main Activities

  • CTO and technical partner to the CEO: own strategy, roadmap and delivery of Divimah's travel CRM and itinerary platform; lead a distributed engineering team through sprint planning, mentoring, code reviews and releases.
  • Full-stack ownership: Angular (Signals, NgRx, RxJS) on Vercel, NestJS/Node.js + MongoDB on DigitalOcean, Redis caching and distributed locking, GitHub Actions CI/CD, QA environment ahead of production.
  • Productized the chat app with Ionic/Capacitor for iOS and Android: WebSocket messaging, presence, scheduled messages and push notifications.
  • Leading the AI roadmap: an ML recommendation service that suggests services while agents build itineraries, and LLM-assisted operations suggestions built from each agency's rules.
  • AI-native engineering: Claude Code and Codex in the daily workflow, with project agent skills (live-database playbook with dry-run guardrails, code review, ticket automation), parallel agents in isolated git worktrees, and browser rigs that reproduce each bug before and after the fix.
  • Replaced Jira with an in-house board (Next.js, Prisma, PostgreSQL with row-level security): sprints, git-to-ticket automation, Discord sync, 1,258 issues migrated.
  • Reliability and data integrity: led the fix and live-data recovery for a long-running client-duplication bug, and fixed memory leaks and an out-of-memory crash in the itinerary builder.

Across the platform, I also drive architecture decisions, lead code reviews, and keep release quality high through clear sprint planning, strong delivery rituals, and pragmatic engineering standards.

Senior Software Engineer
number8/SoftwareMind | Feb 2024 - Jan 2025
HTML/CSS
Python
Node.js
Git
MongoDB
Angular
API
PostgreSQL
Elasticsearch
TypeScript
Sonar
Docker
React
JavaScript
Jest
RxJS
NgRx
API Integration
Figma
UI Development
SCSS
Bootstrap 5
Express.js
AWS
React (next.js)

Main Activities

  • Developed a high-scale data management platform for enterprise documents and personnel workflows, improving retrieval across varied formats (PPT, Excel, PDFs, and other assets) with stable APIs and distributed indexing.
  • Shipped scalable modules with Angular, Elasticsearch, Node.js, and PostgreSQL to improve searchability, operational reporting, and role-based data access for large-volume client portfolios.
  • Modernized user interfaces for responsiveness and performance, including reusable component patterns, robust state handling, and accessibility-aware layouts for desktop/mobile users.
  • Introduced AI/LLM-assisted workflow ideas (prompt-based classification and smart recommendation patterns) to support teams in faster triage, content discovery, and knowledge retrieval tasks.
  • Collaborated closely with product, QA, and delivery stakeholders through Agile ceremonies, and drove feature quality with Azure DevOps, GitLab, and Sonar-based quality governance.

Also contributed directly in React/Next.js where it strengthened UX delivery, supporting faster customer-facing experiences and cleaner frontend integration with backend services.

Projects

AI-native engineering system (Claude Code + Codex)
2025
Python
TypeScript
Godot Engine
Prompt Engineering
LLM
Multi Agent frameworks
Mcp connections
Claude code
Ai agent
Agentic system design
My own agentic development setup, used every day on client and product work. 60+ reusable agent skills (domain playbooks, guarded live-database operations, code review, ticket automation, iOS release flows), MCP integrations wired into the agents' toolset, and a spec-driven multi-agent pipeline: specs are split into work rows, parallel agent teams build them in isolated git worktrees with each task routed to the right model tier, and an adversarial review gate runs before any PR. Verification is headless and automatic (builds proven by exit code, scripted end-to-end app journeys with screenshots), so agents prove their own work. On a Godot 4 mobile game, this pipeline turned 148 specs into 17 build waves, including a single wave of 24 parallel agent teams with 34 of 35 deliverables accepted.
ragzinho: local hybrid RAG over a personal knowledge base
2026
Python
Machine Learning
Information Retrieval
Semantic search
LLM
RAG
Embedding models
Rag Based architectures
Ollama
A local retrieval layer that answers questions over my notes, research reports, project docs, video transcripts and a 365,000-item art library, with nothing leaving the machine. Hybrid search (BM25 + bge-m3 embeddings served by Ollama) combined with reciprocal rank fusion, optional cross-encoder reranking, image vectors for visual search, pluggable corpus adapters, golden-set evaluation before any model swap, and an incremental re-index daemon that keeps every collection fresh. Exposed as a CLI and a warm local server that coding agents query before they search the web.

Education

Universidade UNINOVE
Bachelor's degree・Systems Analysis and Development
Jan 2020 - Dec 2023
ETEC Albert Eistein
Undergraduate Degree in Information Systems・Information Systems
Feb 2009 - Jul 2011

Certifications & Awards

CAPGEMINI BEST AGILE PROJECT
THE NORTH AMERICAN SOFTWARE TESTING AWARDS 2022 | Nov 2022