Backend developer with a strong transition to Applied AI & LLM Systems. From mission-critical SAP ERP (SKANSKA, Banco do Brasil, Claro) → modern backend (Node.js, TypeScript, DDD, Clean Arch, SOLID) → now building production-grade RAG, multi-agent workflows, and fine-tuned models.
Philosophy: AI fails due to poor engineering, not models. I focus on architecture, evaluation (RAGAS), observability, and real impact.
- Theological LangGraph Agent: Multi-agent biblical analysis (parallel execution, LangGraph + FastAPI + Streamlit)
- RAG + RAGAS Evaluation (Insurance/Legal Docs): Agentic RAG with adaptive chains, rerank, HyDE, composite scoring
- Essay Grader (DistilBERT Fine-Tuned)
Regression model for essay scoring (0-10) – MAE 1.09 using HF Trainer & Torch - Background Removal
Instant image background remover using fine-tuned vision model - Simple Chatbot
Basic conversational interface with LLM integration
- theological-langgraph-agent → Multi-agent theological exegesis
- rag-evaluation-contracts-ragas → Advanced RAG with RAGAS metrics for contracts/insurance
- Gym-Pass-Like-API → SOLID + Clean Arch backend
- Payment-Gateway → Full-cycle microservices (Go + NestJS + Kafka)
More: LangGraph • RAGAS • Torch • FastAPI • Streamlit • pgvector/Pinecone • Docker • CI/CD
- 📧 oliiveirap90@gmail.com
- 🔗 LinkedIn: linkedin.com/in/israeloliveira-isop
- 🤗 Hugging Face: huggingface.co/isop90