Open to opportunities

Tendresse Dutra

@tendressedutra

Message

Senior Software Engineer · AI Systems

Italy

Message

What I'm looking for

I'm looking for a full-remote role building ML systems end-to-end — training, serving, and observability. I want to own the full pipeline, not just one layer. Teams that care about engineering rigor in AI: reproducible experiments, proper evaluation, production-grade APIs. The intersection of ML and backend infrastructure is where I work best.

After many years working in different environments and tech stack I evolved toward AI Systems. I integrate AI in the real world production to ensure awareness and reliability.

Experience

Work history, roles, and key accomplishments

Current

AI Engineer – Vision Model Dev

Current

VISUAL-FINETUNE-LAB (GitHub project)

Jan 2026 - Present (5 months)

End-to-End Vision Model Specialization Pipeline. OpenCV preprocessing → GPT-4o Vision synthetic dataset generation → QLoRA fine-tuning on Phi-3.5-Vision (PyTorch + HuggingFace PEFT, 4-bit NF4, bfloat16) → BLEU/ROUGE-L/LLM-as-judge evaluation → FastAPI serving on Azure Container Apps. ~10x cost reduction vs GPT-4o at runtime.

Python OpenCV QLoRA MLFlow fastAPI Docker Azure PyTorch Hugging Face PEFT

AI Engineer LLMOps/Observability

LLM-QA-OPS-LAB (GitHub project)

Nov 2025 - Jan 2026 (2 months)

AIOps intelligence layer for production incident analysis. Redis Streams ingestion with batched LLM calls and self-reported hallucination risk. Rule-based + LLM-as-judge faithfulness evaluation tracked in Prometheus. RAG pipeline with pgvector + LlamaIndex + LangGraph StateGraph with conversation memory. Grafana dashboards auto-provisioned via Docker/Kubernetes.

Python OpenAI API LLM Output Evaluation RAG Pgvector LlamaIndex LangGraph Grafana Prometheus Docker Kubernetes fastAPI Redis

Backend Engineer

Swag International

Aug 2022 - Apr 2025 (2 years 8 months)

Built backend microservices for a fintech/crypto platform using event-driven Kafka/Confluent pipelines to enable real-time sync across 12+ modules. Implemented observability with Elasticsearch to cut mean time to incident detection from hours to minutes and reduced Customer.io email workflow redundancy by 60% using Liquid templating.

Spring Boot Kafka Confluent MySQL Microservices Elasticsearch Event Driven Architecture Observability Java

Backend Developer

Next4B

Mar 2022 - Jul 2022 (4 months)

Developed Python APIs and a MongoDB data layer for financial reporting dashboards, building data aggregation pipelines for client-facing analytics. Monitored production performance with Grafana.

Python REST APIs MongoDB Data Aggregation Financial Reporting Monitoring Grafana Backend Services