Expressive voice technologies for the film and entertainment industry• Leading the Audio AI team in research and development of expressive voice technologies• Utilizing deep learning and digital signal processing to enhance vocal emotion and expressiveness• Awarded the Gründungsstipendium by the government of North Rhine-Westphalia (NRW), a grant which supports innovative business ideas

Audio plugin company building the next generation of neural audio products• Designed and implemented novel deep learning architectures for cross-lingual singing-voice conversion in low-resource languages + generative music models for drums and wind instruments.• Improved MOS scores by 55% through iterative refinement of model architecture and training strategy. Project was covered by BBC and used in movie grossing over ₹850 million.• Integrated deep learning models with JUCE to develop AU/VST3 plugins. Compressed and optimized PyTorch models for realtime inference with pruning, quantization, and distillation.• Deployed generative audio model as a web-based GUI interface hosted on AWS. Set up CI/CD pipeline using GitHub Actions, BentoML, and Docker, saving 15 man-hours per month.• Pitch Innovation's plugins are used by Grammy and Oscar-winner AR Rahman, 4x Grammy nominated & multi-platinum producer Robot Scott (H.E.R, Kanye West, Ty Dolla, BeBe Rexha), Jordan Rudess of Dream Theatre, EDM artist KSHMR, and many others.

Data science firm accelerating decision-making for multinational enterprises• Led team of 4 data analysts delivering projects for international clientsgenerating revenues of up to $25million• Utilized advanced statistical techniques, including Bayesian inference andtime series analysis, to model complex data structures, resulting in a20% reduction in forecasting errors for a demand planning project• Reduced cycle time of routine data migration activities by 66%, throughautomation scripts. Also wrote scripts to automate data ingestionand data entry tasks, saving 200 man hours per month

Privacy-preserving analytics application making Shopify businesses smarter• Building on undergraduate thesis research, created a Python library forpreserving privacy of user data when training machine learning models,by using federated learning and differential privacy• Deployed a comprehensive suite for privacy-preserving analytics andintegrated it with Shopify APIs to serve it as a Shopify app• Successfully acquired B2B users from the e-commerce space across 4different countries on a subscription model• Increased client average order value by 33% and reduced churn by 10%on average, in under 6 months of deployment

Was part of the core AI R&D division at Saama. Ideated and built blockchain prototypes for clients in the clinical trials industry. Also implemented deep learning architectures from research papers.