Senior Full-Stack Engineer and AI Engineer with over a decade building distributed systems and, in recent years, productionizing Generative AI solutions.
I specialize in designing end-to-end architectures that combine robust backend engineering, cloud-native infrastructure, and advanced AI capabilities — turning experimental LLMs into reliable, secure, and scalable enterprise products.
Currently at Ceibal, I lead the architectural design and implementation of AI-infused platforms that serve national-scale educational impact.
- 🌱 Deepening expertise in AI Agents orchestration, evaluative frameworks, and MLOps for LLMs
- 🔭 Passionate about system design, homelabbing, and Open Source
- 💬 Open to discussions on RAG at scale, multi-agent systems, and production AI reliability
Frontier LLMs (GPT-5 series • Claude 4 • Gemini 3 • Llama 4) • Multi-Agent Systems • RAG Pipelines at Scale • Vector Databases • LangChain / LlamaIndex • Prompt Engineering & Evaluation • Fine-tuning & Quantization • Model Orchestration & Routing • AI Observability & Cost Optimization
Senior Full-Stack Engineer | AI Engineer
Ceibal — Montevideo, Uruguay (Dec 2019 - Present)
- Designed and led the architecture of Generative AI solutions using frontier models (GPT-5, Claude 4, Llama 4), implementing advanced RAG pipelines and autonomous agent systems that significantly reduced operational workload.
- Engineered high-performance microservices and distributed architectures optimized for AI workloads, ensuring seamless integration with third-party platforms at national scale.
- Architected intelligent process automation combining BPMS (Camunda) with AI decision engines, delivering 40% efficiency gains in critical workflows.
- Led End-to-End delivery of multiple AI-infused products — from architectural design and requirements translation to secure, scalable, observable production deployments.
- Established robust CI/CD, monitoring, and AI-specific observability pipelines (latency, token usage, hallucination detection, and model drift).
- LinkedIn: Denys Retureta
- GitHub Issues: Start a discussion
"I don't just integrate LLMs — I architect complete, reliable, production-ready AI systems that deliver real business value."
Made with ❤️ and ☕ in Uruguay