Senior Software Engineer · AI Engineering · Security Engineering · Application Security · DevSecOps
Building production software, AI systems and secure platforms with Rust, Python, TypeScript, Java and cloud-native technologies.
Software & AI Engineer @ Dewtech · Belo Horizonte, MG, Brazil (UTC−3) · Remote
I'm a Senior Software Engineer with 10+ years of experience building production systems, with strong specialization in AI Engineering, Application Security, Security Engineering and DevSecOps.
My work spans backend systems, distributed architectures, cloud-native platforms, developer tooling and AI-native applications using Rust, Python, TypeScript/Node.js, Java, .NET, Docker and Kubernetes.
Over the last several years, my work has increasingly combined software engineering with security engineering, including Application Security, Secure SDLC, SAST, DAST, SCA, threat modeling, vulnerability remediation, secure architecture, offensive testing and CI/CD security.
More recently, I have been building and securing LLM applications, AI agents and MCP systems, including agent runtimes, RAG/GraphRAG architectures, runtime protection, automated red teaming, tool authorization, prompt-injection testing and security controls for AI-native systems.
My goal is simple: build software that works in production and design it to be secure from the start.
Technical work spans dewtech-technologies, darelabs-tech and fermio-technologies, covering software engineering, AI platforms, developer tooling, application security and applied AI-security research.
My core background is software engineering. Security is embedded in how I design, build and ship systems rather than added only at the end of the lifecycle.
- Backend & distributed systems — APIs, microservices, queues, event-driven systems and integration layers
- Cloud-native engineering — Docker, Kubernetes, GitOps, CI/CD and production deployment workflows
- Systems programming — Rust and C++ for performance-sensitive tooling, runtimes and infrastructure
- AI engineering — LLM applications, agents, RAG, GraphRAG, tool calling, multi-provider routing and evaluation workflows
- Developer tooling — CLIs, orchestration frameworks, testing harnesses and automation
- Data platforms — PostgreSQL, Redis, Elasticsearch, MongoDB, Neo4j, Qdrant and Kafka
- DARE Agent Coding Runtime — Rust — agent execution runtime and evaluation harness with containerized sandboxes, verification tests and calibrated CI baselines
- DARE Framework / DARE CLI — agentic engineering toolkit with parallel DAG execution, deterministic build/test/lint verification, GraphRAG and MCP integration
- DARE Swarm — research on evolving multi-agent SLM topologies using deterministic engineering fitness functions instead of LLM-as-judge
- DARE Synapse — C++ / CUDA — Transformer architectures, training and inference runtimes built from first principles
- Fermio Platform — multi-tenant AI agents platform using NestJS, Nuxt, FastAPI and multi-provider LLM routing
- Fermio Engineering Copilot — GraphRAG + parametric CAD + FEA/CFD pipelines for mechanical engineering
- Document Quality Scoring API — production rewrite from Python to Rust/Axum/Tokio
Deterministic adversarial validation and security conformance testing for AI agents and MCP systems.
Evidence-first by design: conclusions are backed by reproducible test vectors, deterministic expectations and machine-readable evidence for CI/CD — not by an LLM acting as the final security judge.
Scope includes MCP server/tool discovery, security baseline generation, authorization and policy validation, AuthZEN / COAZ-MCP conformance vectors, mappings to OWASP Agentic Security, controlled adversarial validation and agent/tool/resource attack-path modeling.
→ github.com/darelabs-tech/dare-agent-security
Open-source static analysis CLI written in Rust for PHP codebases such as Laravel, Symfony and WordPress.
Includes taint analysis for command injection, SQL injection and XSS, deterministic findings with stable fingerprints, declarative rulepacks, SARIF output and baselines for CI adoption.
→ github.com/fermio-technologies/fermio-sec-cli
Open-source Model Context Protocol server for YouTube intelligence in Claude Desktop, designed with OAuth2, AES-256-GCM token encryption at rest, SSRF protection, rate limiting, audit logging and Zod-validated inputs.
Listed in the official MCP Registry.
→ tubemind.dewtech.tech · GitHub · npm
Open-source MCP bridge for Obsidian with path-traversal protection, input sanitization, localhost-only enforcement, API-key isolation, size limits and audit logging.
Listed in the official MCP Registry.
Runtime Application Self-Protection for AI workloads in production, with a Rust core and Python/JavaScript instrumentation.
Pentest platform with a human pentester in the loop, focused on automated red teaming against LLM applications, including prompt injection, jailbreak fuzzing and tool-call exploitation.
I work across both traditional application security and emerging AI-native attack surfaces.
- Application Security — SAST, DAST, SCA, secure code review, vulnerability remediation and OWASP Top 10
- Secure SDLC — security requirements, threat modeling, secure architecture and CI/CD security gates
- Vulnerability Management — detection, triage, risk classification, reporting, remediation guidance and follow-up with engineering teams
- Offensive Security — penetration testing, adversarial testing, red teaming, attack-path analysis and exploit-oriented validation
- AI Security — LLM applications, MCP, agentic systems, prompt injection, jailbreak testing, authorization, policy validation, RAG/GraphRAG security and secure tool use
- Runtime Security — RASP, guardrails, policy enforcement, auditability and deterministic controls
- Cloud & Edge Security — Kubernetes, GitOps, Cloudflare, AWS, OCI and GCP
- Enterprise AppSec tooling — GitLab Security Pipelines, Veracode and Checkmarx, with evaluation/POC experience using Snyk and Black Duck
- Software Engineering — backend systems, distributed architectures, APIs, microservices, developer tooling and production systems
- AI Engineering — agents, RAG, GraphRAG, tool calling, structured outputs, multi-provider LLM routing and observability
- AI Security — LLM security, agent security, MCP security, tool authorization, runtime protection and adversarial testing
- Application Security — Secure SDLC, SAST, DAST, SCA, threat modeling, secure code review and remediation
- Security Engineering — automation, policy enforcement, CI/CD security, deterministic validation and auditability
- Systems Engineering — Rust, C++, containerized runtimes, orchestration and performance-oriented tooling
- DevSecOps — Kubernetes, Docker, GitOps, CI/CD, cloud infrastructure and security gates
Languages: Rust · Python · TypeScript · JavaScript · Java · C#/.NET · C++ · PHP · Ruby
Backend & frameworks: Node.js · NestJS · FastAPI · Spring Boot · .NET Core · Laravel · Flask · Ruby on Rails
AI / Agentic: OpenAI · Anthropic · Gemini · MCP · RAG · GraphRAG · LangChain · Hugging Face · Qdrant · Neo4j
Security: SAST · DAST · SCA · Secure Code Review · Threat Modeling · Vulnerability Management · Secure SDLC · OWASP Top 10 · OWASP LLM / Agentic Security · Penetration Testing · Security Gates
AppSec tooling: GitLab Security Pipelines · Veracode · Checkmarx · Cloudflare · Snyk (POC) · Black Duck (POC)
Cloud & platform: Kubernetes · Docker · GitOps · AWS · GCP · Oracle Cloud · Cloudflare · PostgreSQL · Redis · Elasticsearch · MongoDB · Kafka
- DARE Agent Security — adversarial validation and security conformance for AI agents and MCP systems
- Fermio Sec CLI — Rust static analysis and taint-tracking CLI
- tubemind-secure-mcp — production MCP server with security-by-design controls
- obsidian-mcp-secure — secure local MCP integration
- DARE CLI — Rust-native agentic engineering CLI
Open to selected opportunities in:
Senior Software Engineering · AI Engineering · AI Platform Engineering · Security Engineering · Application Security · Product Security · AI Security
- 📧 wleandro.oliveira@gmail.com
- 🌎 Belo Horizonte, MG, Brazil (UTC-3 / BRT) · Remote
- 💼 Software, AI & Security Engineering @ Dewtech