Teanna Sims Email & Phone Number

United States

United States

United States

Developed an AI party DJ using pose estimation for crowd behavior analysis and GNN data interpretation. Enhanced AI performance with multiple encoders and completed training in Intel Developer Cloud for further skill development.• Applied GNN for crowd movement analysis• Implemented multiple encoders for improved AI performance• Completed training in Intel Developer Cloud

• Developed a highly adaptable LLM-based agent architecture for multi-agent cooperation using Cerebras inference.• Achieved significant speedup in processing large multimodal datasets compared to traditional GPU-based systems.• Collaborated with cross-functional teams to optimize system performance and enhance overall efficiency.

• Designed an open-source multimodal masking pipeline using synthetic data generation and pseudonymization to enhance AI model fine-tuning.• Integrated Apache Spark with BabitMF for high-performance video transcoding, reducing preprocessing time significantly.• Developed a parallel synthetic data generation pipeline incorporating video dense captioning and keypoint estimation for improved model accuracy.• Engineered audio transcription and synchronization for realistic synthetic video outputs, presenting at ByteDance workshops and Developer Summit.

• Collaborated with interdisciplinary teams to design and deploy VR games and simulations at the University of Michigan.• Developed best practices for XR project development, ensuring efficient planning, execution, and evaluation processes.

Redesigned and modernized a trading market simulator, transitioning it from Java to Python for improved efficiency and functionality.• Enhanced the simulator's efficiency and functionality for more accurate testing and analysis.• Researched and developed deep learning models to test implications of effective deep hedging models in real-world markets.

New York, New York, United States

• Remodeled Buhler’s Deep Hedging framework using Imaki’s no-transaction band network for optimal hedging strategies.• Developed a gamma volatility backtesting model using Ornstein-Uhlenbeck model for volatility predictions.• Utilized neural network architecture to identify optimal hedging strategies in incomplete markets.

• Analyzed large emissions datasets to build emission comparison models for diesel and solar powered motorcycles.• Co-authored academic publication on 'Emissions impacts of electrifying motorcycle taxis in Kampala, Uganda.'• Contributed to research published in Transportation Research Part D: Transport and Environment.

Boston, Massachusetts, United States

LSA - Comprehensive Studies Program Course: Math 105

For more information about optiMize's incubator program visit https://www.optimizemi.org/sic

Data Science

Data Science

Activities and Societies: optiMize, NSBE, Delta Sigma Theta, B3