• Stayed ahead in the deep learning field by incorporating cutting-edge research and methodologies into our product offerings. Ensured our company remained a market leader, leveraging our agility and innovation despite having a smaller team and limited resources.• Developed an image search capability, akin to Google Lens, enabling customers to instantly locate similar objects within their datasets, enhancing user experience.• Directed the integration of a large language model (LLM), broadening the scope of the company's ML applications.• Led efforts in computational efficiency, implementing optimisations across the stack to reduce resource consumptionBusiness Outcomes• Achieved significant cost savings and improved service delivery through model and pipeline optimisations, directly contributing to the company’s competitive advantage and financial performance

• Pioneered the development and deployment of the company's deep learning infrastructure, comprising 5 specialised ML models.• Analysed millions of high-resolution images, achieving industry-leading false positive rates in anomaly detection.• Developed ETL pipelines for large volume data, resulting in more than 1 billion inferences made with the machine learning models.• Collaborated closely with customers to understand their data analysis needs, facilitating the translation of complex data insights into actionable business strategies• Leading code contributor across stack, utilising PyTorch, Python, and DockerBusiness Outcomes• Enabled the company to secure substantial yearly service revenue through the deep learning suite, setting new benchmarks within the industry.• Surpassed customer expectations in precision and recall metrics, delivering critical analysis to both government and private sector clients.

Cartographic applications using Machine Learning• Led the design and engineering of a custom U-Net model in collaboration with ETH Zurich toperform Swiss-style relief/terrain shading.• Managed high volume hyper-parameter tuning through Weights & Biases agents.• The team's innovative work was integrated into the recognised Natural Earth dataset, NationalGeographic maps, and resulted in three publications: IEEE 2020, AutoCarto 2022, and NACIS2022.Contributed to the engineering of Eduard, a macOS application for Cartographers, developedby a two-person team, featuring research-driven innovations: http://dilpreet.co/projects/eduardCo-supervised the theses of two Master's students, focusing on Machine Learning applicationsin Cartography.

Led the development of innovative machine learning applications, from healthcare analytics to interactive art platforms. My work, blending research and technical skills, resulted in impactful projects and presentations at prestigious conferences like NeurIPS and EvoStar, contributing to both academic and professional communities.Interactive Augmented Surfaces• Developed an overhead tracking and projection system to transform any surface into an interactive workspace, using a cost-effective and portable design.• Successfully deployed this technology in two major projects: enhancing fluid dynamic simulations for Woodside FutureLab at Monash and creating an engaging Interactive Renewable Energy Table for Monash Tech School.Healthcare and Machine Learning• Led the integration of Machine Learning into the 'Smart Shoe’—a collaborative project withMonash Health and Blundstone.• Developed an iOS app to analyse gait variations using pressure sensor data, implemented withSwift for on-device inference and Python + PyTorch for cloud-based model training.Art and Machine Learning• Created an application for Similarity Search using CNN models; applied this to a scraped andassembled public domain artwork repository (100,000+ artworks).• This concept was expanded into multiple applications, including an iOS app for interactive artexploration, presented at NeurIPS 2018, and a macOS app that merged generative art withsimilarity search, published at EvoStar 2019.

Part of the Advanced Security Center team.