About Isaac

Data & AI Engineer focused on reliable LLM applications, evaluation workflows, and backend AI systems. Undergoing a Bachelor of Engineering in Data and AI Engineering.

Professional Experience

July 2024 - Present Cali, Colombia (Contract)

AI & Web Systems Engineer — Sky Aviation Club

  • Owned Linux-hosted web systems for an aviation training organization, managing deployments, server configuration, troubleshooting, and security hardening.
  • Built backend scripts and workflow automations for document-heavy operational tasks, reducing manual administration through pragmatic internal tooling.
  • Translated stakeholder requests into shipped platform improvements by scoping small releases, validating behavior, and iterating directly with administrators.

Selected AI & Data Projects

2026 SkyTalkTrainer

Aviation Training AI Backend

FastAPI, WebSockets, Local LLMs, Evaluation Harnesses, Guardrails

  • Built a real-time aviation training backend, orchestrating speech/audio pipeline stages with local OpenAI-compatible endpoints.
  • Designed an ICAO ATC evaluation pack covering normal, negative-readback, and noisy-STT scenarios.
  • Implemented guardrail and regression tests for controller intent, phraseology, and unsafe output prevention.
2025 SaberPro AI Assistant

Retrieval-Augmented Guidance System

FastAPI, ChromaDB, Gemini, RAG, WebSockets, Docker

  • Built a RAG assistant by ingesting official Saber Pro Q&A content into persistent ChromaDB collections using multilingual-e5 embeddings.
  • Improved retrieval reliability with hybrid semantic search and lexical boosting for domain terms (PRISMA, ICFES).
  • Delivered real-time AI chat with prompt augmentation, Gemini streaming, input sanitization, and cooldown logic.
  • Created a benchmarking loop measuring precision@2, recall@2, MRR@2, BLEU, and ROUGE-L.
2026 Cloud Lakehouse Platform

Medallion-Style Data Lakehouse

Spark, Airflow, Iceberg, Nessie, MinIO, Trino, Dremio, Azure

  • Built a medallion-style lakehouse, orchestrating Bronze-to-Silver-to-Gold pipelines across MinIO, Iceberg, Nessie, Trino, and ClickHouse.
  • Implemented DLT/PyArrow Bronze ingestion with 250K-row batches and SparkSubmitOperator jobs for Silver cleansing.
  • Extended data movement with Azure Blob export notebooks using DLT, PyIceberg, PyArrow, and ADLFS.

Education

Expected May 2026 Cali, Colombia

Universidad Autonoma de Occidente

Bachelor of Engineering in Data and AI Engineering

Coursework: AI Systems, Data Engineering Pipelines, Machine Learning, Neural Networks, Cloud Infrastructure, Software Engineering, Big Data Frameworks.