Karan Shah

Researching and operationalizing the construction of a digital twin of the city of Ingolstadt, BY, DE using ArcGIS CityEngine and Unity3D. Running procedural tests on a landscape of approx 30 sq. km. with an aim to answer questions in the domains of traffic efficiency, safety, ecology and social acceptance of autonomous vehicles.

Modelled Rabi and Electron Spin Resonance (ESR) data using techniques like Gaussian Process, CMA-ES and a custom Auto-Encoder to better understand and streamline the process of creation of Diamonds with Nitrogen Vacancies and their properties.

Worked on handling customer feedback processing as a Big Data application within the Global Customer and Dealer Satisfaction Team at BMW. Prototyped a Document Clustering pipeline using SBERT and Doc2Vec. Led the process management pipeline during the market launch of the new-7 series (2022) in the European and North American Markets.

Learnt local & global navigation behaviors via multimodal interaction between agent-agent and agent-environment in real world 3D environments using Unity ML Agents and Open AI Gym. Matched benchmark OpenAI benchmarks, and devised connectors to run OpenAI algorithms on packaged Unity environments.

Worked with a decision-science team on multiple use cases in Machine Learning & Computer Vision. Led decision-making conversations using market trends, user behavior findings, price & market constraints and data visualizations.