Camilo Pestana

Joined as Data Science Manager leading a 13-person Data Science & Analytics team, and promoted within 18 months to lead the central AI/ML & Data Science function. In a federated model where each team owns its own delivery, I set direction through the AI council, governance, policy and security.

• Built and shipped fraud detection combining Agentic AI, LLMs and Computer Vision on AWS, generating $230M+ ROI in year one.
• Productionised agentic AI on AWS Bedrock with Guardrails and LangGraph orchestration, moving GenAI from prototype to governed enterprise operations.
• Leading enterprise modernisation from legacy Databricks to AWS (SageMaker, Bedrock, S3 Tables & Apache Iceberg, MSK streaming), cutting cost while improving reliability.
• Defined AI governance standards; AI-security SME for the central security function (threat modelling, red-teaming, defence for agentic and ML systems).
• Shipped a production fraud-detection model for the United Slots launch in 4 weeks, removing initial fraud exposure and protecting revenue from day one.
• Partner to the CTO on the AI/ML roadmap and board-level strategic risk.

Led ML model development and MLOps for clinical AI products.

• Contributed to two FDA 510(k)-cleared models (stenosis estimation and plaque classification) through rigorous clinical validation and regulatory submission.
• Designed and trained 3D CNN attention architectures (PyTorch/TensorFlow) for coronary CT analysis, significantly outperforming baselines.
• Clinical MLOps on AWS SageMaker with MLflow for regulatory-grade experiment tracking and reproducibility.

Joined as Senior Data Scientist (Tech Lead); promoted to Head within 9 months. Built the entire data, software and ML capability from zero; led 7 direct reports.

• Built the company's first data platform from scratch: data lake, warehouse and all AWS pipelines (Airflow, Lambda, Aurora/PostgreSQL).
• Biosensor ML for continuous lactate monitoring; accelerated R&D insights by 50% fortnightly. $65K saved in the first 4 months.

Dual role: Cloud Computing lecturer (2019 to 2021) and DARPA-funded adversarial ML researcher.

• US DoD / DARPA-funded program on adversarial ML: novel attack and defence strategies for computer vision systems.
• Published in IEEE CVPR, WACV and 3DV. Concurrent Chevron ML contract (Oil & Gas, 2020 to 2021).

Co-founded a CV/ML startup building real-time hazard detection for road infrastructure. Led all AI development and product direction from inception.

• Deep CNNs for real-time roadside hazard detection from video, in partnership with Mainroads WA.
• 1st place at the WA Ministry of Data Hackathon (40+ teams); secured a $21K SPURonWA innovation grant within five months.