AI Engineer building production systems that ship and stay running.
Python, AWS, Terraform. Event-driven serverless infrastructure, LLM applications, and the architecture work underneath both.
When does agency earn its cost? — benchmarking LLM extraction strategies on legal documents
Same documents, same schema, three model families. The agent rarely beats a single LLM call. A clean counterexample to the reflex of wrapping every task in an agent loop.
LLM Engineering · Agents · Evals
agentic-kie — schema-driven key information extraction from documents
Typed Python library where a PDF enters as a file path and leaves as a validated Pydantic instance. Handles text-layer detection, OCR routing, image rendering, LLM orchestration, and retry logic. Two extraction strategies — single-pass and agentic ReAct — satisfy the same protocol and swap without touching downstream code. Model-agnostic via LangChain's BaseChatModel.
Python · LangChain · Pydantic · ReAct
agentic-kie-deploy — production-grade serverless AWS pipeline
Terraform-managed event-driven AWS infrastructure for the agentic-kie library. Pre-signed S3 uploads trigger an event-driven Lambda extractor via EventBridge and SQS, writing structured results to DynamoDB. Environment-scoped IAM, plan-bound prod deployments, and ADR-driven architecture decisions.
Terraform · AWS · Lambda · EventBridge · SQS · DynamoDB
Strict typing and high coverage as defaults, not aspirations.
📬 Writing about Agentic AI, LLM Engineering, and ML Engineering → gabriel.com.gt/blog