Alex DE SOUZA
Chief Technology Officer
Chief Technology Officer
Open to senior AI & full-stack engineering workLLM systems · multi-agent orchestration · B2B SaaS
B2B AI SaaS · CTO
Multi-tenant B2B SaaS for the Brazilian real-estate market, architected end to end (Next.js App Router, tRPC, Prisma, PostgreSQL). I designed a LangGraph + Claude agent layer spanning 11 task types for lead analysis, response generation, and campaign automation.
LLM research · Python
Causal Rationale Attribution Faithfulness Test — an open-source library that empirically checks whether an LLM agent’s stated reasoning actually drove its decision, using a randomized perturbation experiment with a placebo control. Zero runtime deps, mypy-clean, CI-ready CLI.
AI safety · TypeScript
Adversarial stress-test harness that injects YAML-defined attacks into tool-using LLM agents and uses a Claude LLM-judge to classify compromise vs. safe refusal vs. detection. Covers 3 victim models, 5 attack categories, and 4 safety metrics, persisted via Prisma with a Next.js dashboard.
Dev tooling · Python
Production CLI/library that runs AI coding agents (Claude Code, Aider, GPT-4o, Gemini) inside isolated Docker containers backed by git worktrees, then fast-forward merges the results. asyncio + Pydantic v2 + Typer, 5 backends, 48 passing tests.
Multi-agent security · Python
Multi-agent SIEM testing whether LangGraph-orchestrated agents beat a monolithic LLM on threat-detection accuracy and cost. Ingests real intrusion datasets (CIC-IDS2017, UNSW-NB15), streams events over Redis, and gates response actions behind a tRPC human-approval UI.
Product & architecture
I turn unclear product ambition into system architecture and a shipping cadence that compounds. At ImpulsoLead I architected a multi-tenant B2B SaaS end to end and grew into the CTO seat.
Full-stack systems
Next.js and tRPC on the front, FastAPI and async pipelines on BullMQ + Redis behind them. I care about speed, maintainability, and leverage in equal measure.
AI & LLM systems
I design multi-agent orchestration with LangGraph and the Claude SDK, RAG with pgvector, and the evaluation harnesses — faithfulness, safety, red-teaming — that keep them honest in production.
Moved quickly from fundamentals into real builds, learning by putting work under pressure instead of leaving it in tutorial mode.
Solo-delivered websites, full-stack apps, and REST APIs for local businesses and research projects across TypeScript, Python, PHP, and C# — from requirements through deployment and maintenance.
Technical higher education (CTeSP) at ESTGA — systems & network programming, network security, algorithms, discrete mathematics, and databases.
Architected a multi-tenant B2B SaaS for the Brazilian real-estate market end to end — Next.js App Router, tRPC, Prisma, and PostgreSQL.
Owning architecture, AI strategy, and infrastructure — including a LangGraph + Claude agent orchestration layer spanning 11 task types, on an async BullMQ + Redis pipeline.
Professional certifications across AI, security, and systems
AI agent task types orchestrated in production at ImpulsoLead
Projects shipped across web, AI, ML, and systems
Enquiries welcome for
Clear scope beats vague hype. Send context and I will read it properly.
Aveiro, Portugal · replies within two days