I'm a Machine Learning Engineer and Platform Architect passionate about building AI systems that are scalable, reliable, and production-ready.
Over 13+ years, I've worked across AI Engineering, Backend Architecture, and Distributed Systems, helping companies bring real AI products to life.

I work at the intersection of LLMs, infrastructure and engineering, building systems that turn models into real products.

• Retrieval-Augmented Generation (RAG)
• Evaluation & Observability for LLMs
• Vector databases, embeddings pipelines
• Fine-tuning and model specialization
• Prompt engineering and safety layers

• Fast, resilient service layers for AI workloads
• Caching, batching, multiplexing, and throughput optimization
• Distributed architectures for inference + data processing
• Secrets, config, multi-environment orchestration

• High-performance APIs (FastAPI / Python)
• Event-driven systems with Kafka
• Storage, caching, and consistency patterns
• Observability (OpenTelemetry, Prometheus)
• Robustness: retries, fallbacks, circuit breakers, CQRS, outbox

See the docs

I'm building Athomic, a modern backend framework specialized for AI-native and data-intensive applications.

Focus areas include:

• LLM pipelines (streaming, tools, RAG, workflows)
• First-class observability for AI systems
• Resilient messaging + consistency guarantees
• Polyglot persistence for AI workloads
• Extensible plugin architecture

Public release coming soon.

I maintain the Open Omni-Cloud initiative as a research project into verifiable cloud-agnostic architectures, similar to a "TCK for multi-cloud".

• ⭐ Contribute: https://github.com/open-omni-cloud/tck-py
• 📖 Manifesto: https://github.com/open-omni-cloud/tck-py/blob/main/MANIFESTO.md

Languages:
Python, TypeScript

AI/ML:
OpenAI SDK, Vertex AI, Hugging Face, LangChain, LangGraph, FAISS, Chroma, Transformers, Ollama

Frameworks & Systems:
FastAPI, Kafka, Pydantic, Dynaconf, Redis, MongoDB, PostgreSQL, Consul

Cloud & DevOps:
Docker, Kubernetes, Terraform, GitHub Actions, GCP, AWS

Observability & Reliability:
OpenTelemetry, Prometheus, Grafana, aiobreaker, tenacity

Designed for AI-native workloads, with a focus on reliability, observability, and distributed patterns. An internal development platform with:

• secrets system
• messaging abstractions
• resilience primitives
• observability and lifecycle orchestration

Specification + test suite for cloud-agnostic runtime guarantees.

• 💼 LinkedIn: https://www.linkedin.com/in/guandaline/

_ “Architecture is the art of structuring complexity, and software engineering is its most expressive canvas.”_ — Valter Hugo Guandaline