Eduardo Pavez

Founding AI Engineer with 4+ years building generative AI systems and cost-efficient cloud infrastructure; delivered agentic LLM workflows achieving 91% output faithfulness and redesigned infrastructure to cut GPU cloud costs by 93% while increasing media generation throughput 11×. Experienced in LangGraph, RAG, LLMOps and automating end-to-end AI media pipelines to scale production and reduce operating costs.

Founding AI Engineer at Growthfuence AI (2025-05 – Present)

Built and operated an AI-driven social media content pipeline producing 200+ AI-generated posts across different accounts during the first year of the startup.

• Built and operated an AI-driven social media content pipeline producing 200+ AI-generated posts across different accounts during the first year of the startup.
• Designed a LangGraph agentic LLM workflow for automated short-form video script generation with RAG retrieval, achieving 91% script faithfulness and improving template adherence by 49%.
• Reduced GPU cloud costs by 93% while accelerating AI media generation 11× by redesigning an open source Terraform/Packer AWS infrastructure stack and by developing a region-selection tool for spot instances.
• Developed Python orchestration tooling integrating the OpenAI API with a cloud rendering pipeline to automate end-to-end AI media generation, achieving a throughput of 30 images per hour.
• Deployed a near-zero-cost serverless microservice on AWS (Lambda, S3, EventBridge, SNS) for automated daily social media publishing, with multi-account support and proactive alerting for failures and credential expiration.
• Led onboarding and training of two non-technical team members into the content production workflow, enabling full delegation of daily content operations.

Software Engineer at Avature (2024-03 – 2025-05)

Increased observability platform reliability and standardized database provisioning.

• Increased observability platform reliability by migrating Grafana from a Puppet-managed EC2 instance to a multi-pod Kubernetes deployment using Kustomize.
• Built the company's first reusable Terraform module for AWS RDS to standardize database provisioning and enable Grafana migration from SQLite to managed PostgreSQL.
• Raised error handling coverage for v1 API endpoints to 100% by refactoring the PHP backend into an MVC architecture.

AI Research Intern at Intelligent Systems Laboratory (LABSIN) (2021-11 – 2022-12)

Reduced feature computation time and improved ML experiment reproducibility.

• Reduced feature computation time by 75.6% by utilizing the Graph-tool Python library.
• Built Docker containers for feature extraction and ML experiments improving experiment reproducibility.
• Designed and ran baseline scikit-learn experiments with k-fold cross-validation to benchmark with traditional ML algorithms.

Bachelor of Science in Mechatronics Engineering – Universidad Nacional de Cuyo (2025-04)