About Me - Thiago Peraro
About Me
Tech lead, AI researcher and entrepreneur
I'm Thiago Peraro, an AI researcher and tech lead focused on natural language processing and multi-agent systems. My journey blends 4+ years of back-end development (Java/Spring), applied academic research, and product-level technical leadership. Computer Science undergrad at UFG (2019–present, ~83% complete). Currently sole tech lead at Count on Sheep (US crypto tax consulting), NLP researcher at CEIA-UFG on the Koru project, and CNPq fellow on the Sobrevidas project. Founded and operated Clivia (Mar 2024 – Dec 2025). Open to new opportunities in AI Engineering.
Key Achievements
Important milestones in my career
Sobrevidas Project — National Award
I lead the AI team of Sobrevidas — an undergraduate research project at UFG, with a CNPq scholarship — building a SUS-integrated chatbot for oral cancer screening. Our work received the best paper award in Brazil at SBCAS 2025, recognizing the project’s social impact.
Clivia — Own Startup (Mar 2024 – Dec 2025)
Founded and operated Clivia, a multi-agent LLM platform for medical clinics (intelligent SDR via WhatsApp). Architected the full stack from scratch — FastAPI, LangGraph Supervisor, LiteLLM, Redis checkpointer, Docker Swarm with 40+ containers. Closed in Dec 2025 when joining Count on Sheep.
Community Leadership
I co-founded Magnatas.hub in 2023 and act as a community builder at Go.IAs, promoting events and networking in AI. These initiatives help connect professionals and foster an innovation ecosystem in Goiânia.
1st AI Brasil Hero — Award (2026)
Recognized by the Brazilian AI community at AI Brasil / CEIA Open Day after speaking on Second Brain + AI OS, hyper-productivity and agentic coding. Award presented by Juliano Kimura.
Technical Experience
Tech stack and areas of expertise
Back-end Development: • Java & Spring Boot — 4+ years; Flyway (60+ migrations), RabbitMQ, Spring Security OAuth2 (GoiásPrev + Count on Sheep) • Python — FastAPI, Pydantic v2, asyncio • SQL — PostgreSQL, Supabase, pgvector • RESTful APIs, microservices architecture, Linux
AI and LLMOps: • Multi-agent systems — LangGraph (Supervisor, StateGraph, RedisSaver and AsyncPostgresSaver checkpointers), CrewAI, LangChain (LCEL) • MCP (Model Context Protocol) — custom client and FastMCP server implemented from scratch • RAG / GraphRAG — pgvector, FAISS, Supabase vector store • LLMOps — LiteLLM (self-hosted proxy with router and fallback), OpenLit, Helicone, Langfuse (tracing + llm-as-a-judge Evaluator) • Red Team and Guardrails — DeepTeam (prompt injection, adversarial), Deepeval (toxicity, bias, hate speech) • Real-time STT — AWS Transcribe Streaming (WebSocket state machine, KWS sidecar), Deepgram Nova-3, OpenAI Whisper • Custom Text2SQL / GenBI, local Ollama
Infrastructure and DevOps: • Docker Swarm — 40+ containers in production (Clivia) • CI/CD — GitHub Actions → Docker Hub → Portainer redeploy • Observability — Langfuse, LangSmith, Prometheus, Grafana, Loki
Entrepreneurship & Leadership: • Founded, operated, and closed an LLM startup (Clivia) • Community building — Go.IAs (500+ members) and Magnatas.hub • Regular speaker on AI and entrepreneurship
Education & Research
Academic and scientific journey
Academic background • BSc in Computer Science — UFG (2019–present, ~83% complete) • Active participation in research and extension initiatives
Current research (two distinct projects) • Sobrevidas Project — CNPq Fellow (2024–present) — NLP applied to healthcare; a SUS-integrated chatbot for oral cancer screening. Nationally awarded at SBCAS 2025. • Koru Project — NLP Researcher at CEIA-UFG/AKCIT (Dec 2024–present) — LangGraph multi-agent system for assessing soft skills and leadership competencies. I develop the Koru dashboard (Streamlit + Plotly + TimescaleDB) and run adversarial Red Team tests with DeepTeam.
Publications & recognition • 🏆 Co-author of the best tools paper in Brazil at SBCAS 2025 — Sobrevidas Project • 🌟 1st AI Brasil Hero (2026) — recognition from the Brazilian AI community at AI Brasil / CEIA Open Day • Highlighted leader among 60 fellows in the EMBRAPII program, supported by eight PhD mentors • 🎤 Speaker at AI Brasil 2026 — "Second Brain + AI OS" (CEIA Open Day / Go.IAs Experience); regular speaker on AI and entrepreneurship
Philosophy & Vision
How I see the future of technology
I believe in AI’s transformative power when applied ethically and responsibly. My work always seeks balance between technological innovation and positive social impact.
In Sobrevidas, for example, we built a solution that can save lives by enabling early diagnosis of oral cancer. At Clivia (Mar 2024 – Dec 2025), I built production-grade multi-agent tools that humanized medical care through technology.
As a community leader, I promote networking and collaboration, believing innovation emerges from knowledge exchange and shared experiences. I advocate for ethical AI and building technology that benefits society.
Detailed Technical Evolution
Interactive skills timeline by year and category
Explore my technical journey interactively — see which skills I acquired each year, the associated projects, and the professional context behind each technology.