Pestana AI Consulting

AI strategy &engineeringthat delivers.

From boardroom strategy to production-grade AI systems.

PhD-backed expertise. $500M+ in delivered ROI. Defense, diagnostics, fraud, and security.

View servicesBook a consultation
PhD in AIIEEE PublishedDARPA Research15+ Years
$500M+
Client ROI Delivered
2
FDA-Cleared AI Models
15
Years in Tech
20
Person Org Built
5
IEEE Publications

Services

What I deliver

End-to-end AI consulting, from executive strategy to hands-on engineering. The rare consultant who can present to your board and personally build the systems.

AI Strategy & Roadmap

Define your AI vision, identify high-ROI opportunities, and build a phased execution plan. From board-level strategy to team structure and vendor evaluation.

  • AI maturity assessment
  • Use-case prioritisation matrix
  • Phased roadmap with milestones
  • Build-vs-buy analysis

AI/ML Engineering

Hands-on architecture and delivery of production ML systems. Model development, MLOps pipelines, and scalable inference. Not PowerPoint engineering.

  • Model architecture & training
  • MLOps pipeline (CI/CD, monitoring)
  • Cloud infrastructure (AWS/GCP)
  • Production deployment & scaling

Agentic AI & LLM Systems

Design and build autonomous AI agents, RAG pipelines, and LLM orchestration layers. From proof-of-concept to production-grade reliability.

  • Agent architecture design
  • RAG pipeline implementation
  • LLM evaluation & selection
  • Guardrails & safety layers

Computer Vision

End-to-end CV solutions: from 3D CNNs and medical imaging to real-time detection and video analytics systems.

  • Model development (detection, segmentation)
  • Medical imaging pipelines
  • Video analytics & real-time inference
  • Edge deployment optimisation

AI Security & Red Teaming

Adversarial ML assessments, LLM red teaming, and defence implementation. Rooted in DARPA-funded research and Hack The Box credentials.

  • LLM vulnerability assessment
  • Adversarial robustness testing
  • Prompt injection defence
  • Security architecture review

Team Building & Uplift

Build or upskill your AI/ML team. Hiring strategies, capability assessments, training programs, and hands-on mentoring. Scaled a 20-person Data Science, ML, and BI team within a 1,500+ employee organisation.

  • Team capability assessment
  • Hiring strategy & interview design
  • Custom training programs
  • Ongoing technical mentoring
Discuss your project →

Case Studies

Proven results, not promises

Every engagement is measured by business outcomes. Here are three that demonstrate the range.

VGW
01

Fraud Detection System

$4.15B USD revenue

$230M+ ROI in year one

Built and shipped a production fraud detection system combining Agentic AI, LLMs, and Computer Vision for a major gaming platform. The system eliminated initial fraud exposure during a new game launch and continues to protect revenue at scale.

“Personally architected core systems: model architecture, LLM orchestration layer, and CV pipeline. This was hands-on work, not just team management.”

PyTorchLangGraphAWS BedrockSageMakerComputer Vision
Artrya
02

FDA-Cleared Medical Imaging

MedTech, AI for Coronary Artery Disease

2 FDA 510(k) cleared AI models

Led ML model development and MLOps for clinical AI products analyzing coronary CT scans. Two of the models developed received FDA 510(k) clearance for commercial use: stenosis estimation and plaque classification.

“Regulated, high-stakes environment requiring rigorous clinical validation, regulatory submission work, and diagnostic accuracy that directly impacts patient outcomes.”

PyTorchTensorFlowSageMakerMLflow3D CNNs
Open Source
03

AI Security Educational Series

From DARPA research to public education

12 comprehensive modules

A comprehensive open-source course covering the full AI attack surface: adversarial attacks on CNNs, audio attacks, LLM security (prompt injection, jailbreaking, guardrail bypass), model-level attacks (DoS, model stealing, embedding inversion), and data poisoning.

“Bridges PhD-era adversarial ML research with practical, hands-on security education. All modules use local models, no API costs.”

PyTorchDockerAdversarial MLLLM SecurityRed Teaming
View on GitHub →

Process

How I work

Every engagement follows a structured process, from understanding your problem to ensuring your team can own the solution.

1

Discovery

Deep-dive into your business context, data landscape, and AI maturity. Identify high-ROI opportunities and define success metrics.

2

Assessment

Technical feasibility analysis, data readiness audit, and architecture review. Honest evaluation of what AI can and cannot solve.

3

Implementation

Hands-on engineering from model development to production deployment. MLOps pipelines, monitoring, and iterative improvement.

4

Support

Knowledge transfer, team upskilling, and ongoing advisory. Ensuring your team can own and evolve the systems independently.

Credentials

Why clients trust me

Not just theory. Every credential was earned solving real problems in high-stakes environments.

PhD in AI, University of Western Australia

ACS Medal Finalist. Top 10 PhD Thesis in Western Australia.

DARPA-Funded Research

US DoD adversarial ML and explainability program.

Australian Global Talent Visa

Granted for outstanding expertise in AI and Data Science.

5 IEEE Peer-Reviewed Publications

CVPR, WACV, 3DV, IJCNN (15-25% acceptance rates).

Publications

IEEE CVPR2021Assistive Signals for Deep Neural Network Classifiers↗IEEE 3DV2022Attacking Image Classifiers to Generate 3D Textures↗IEEE WACV2022Transferable 3D Adversarial Textures using End-to-end Optimization↗IEEE WACV2021Defense-friendly Images in Adversarial Attacks: Dataset and Metrics for Perturbation Difficulty↗IEEE IJCNN2021Adversarial Attacks and Defense on Deep Learning Classification Models using YCbCr↗

Certifications & Recognition

AI Red Teamer Path

2nd in Australia. 41st globally (796 teams). <43% completion rate. Developed with Google's Secure AI Framework.

Global Cyber Skills CTF: Operation Blackout2025
Agentic AI (7 courses)2025
RSNA Medical Imaging AI (Advanced)2023
ML in Production2022

Awards

2025

Exceptional Impact Award

2025

HTB CTF: 2nd in Australia / 41st Globally

2022

Australian Global Talent Visa

2022

ACS 1962 Medal Finalist

About

The person behind
the consultancy

15+
Years in Tech
PhD
in AI (UWA)
5
IEEE Papers

I'm Camilo Pestana, a Perth (Australia)-based AI leader with 15 years in tech. I started as a software engineer in 2011 and pivoted to Machine Learning in 2017, years before AI became mainstream.

Currently Head of Data Science at VGW ($4.15B revenue), leading a ~20 person AI/ML Engineering & Data Science org. Previously: DARPA-funded adversarial ML research, medical imaging AI at Artrya, and a PhD in AI from the University of Western Australia.

I consult because most AI initiatives fail from a gap between strategy and execution. I bridge that gap. I can define your AI roadmap in the morning and review your model architecture in the afternoon.

Perth, AustraliaGlobal Talent Visa

Domains of Impact

Gaming

$230M+ ROI fraud detection. Agentic AI, LLMs, Computer Vision at $4.15B scale

Defense

DARPA-funded adversarial ML research. Secure systems for military intelligence

Medical

2 FDA 510(k) cleared AI models. 3D CNN cardiac imaging. Biosensor ML

Academia

PhD in AI (UWA). 5 IEEE publications. Cloud Computing Lecturer

AI Security

HTB AI Red Teamer. 2nd in Australia CTF. Open-source security course

Oil & Gas

ML curriculum for geoscience. Upskilled 12 engineers at Chevron

Software

Enterprise software consulting for Southwest Airlines, National Geographic, ESPN, Disney, and Royal Caribbean via Globant

Career path

2011First Engineering Role
2014–2017Globant: Global Scale
2017Pivoted to Machine Learning
2017–2019Co-founded SmartVision AI
2019–2022PhD + DARPA Research + Lecturer
2020COVID-19 AI Detection
2022–2024Artrya: FDA-Cleared AI
2024–NowVGW: Head of Data Science

Get Started

Let's solve your
AI challenge

Whether you need a strategy review, a production system built, or a team upskilled, I'll give you an honest assessment and a clear path forward.

Currently available for Q3 2026 engagements
Book a consultationcamilo@pestana.ai
LinkedInGitHub

Or send a message and I'll respond within 24 hours.