Ayush Garla

Implemented object detection to support smart mobility and sustainable transport systems. Improved traffic sign recognition training efficiency by 60%. Enhanced model performance and cost-efficiency on Google Cloud, boosting processing speed by 64%. Built a custom CNN model for benchmarking and proactively identified performance improvements. Collaborated with cross-functional teams to prototype AI-powered mobility solutions.

Built end-to-end ML data pipelines using AWS S3, Snowflake, and dbt to support computer vision, predictive modeling, and analytical workloads. Developed and validated training datasets for YOLOv8 and CNN models, achieving 99% detection accuracy through optimized preprocessing, augmentation, and labeling workflows. Implemented automated ETL workflows using AWS Glue and Python, resulting in 60% faster data processing and 35% fewer pipeline failures for ML model training. Created transformation models, seeds, and data quality tests in dbt to ensure reliable, high-quality data for ML research. Collaborated with mobility research teams to translate business problems into model-ready data sets, feature engineering strategies, and scalable ML workflows.

Automated data pipelines with Python/PostgreSQL, reducing manual effort by 70%. Developed ETL workflows and scheduled them on VMs to streamline recurring processes. Built regression models to calculate 2000+ employees' performance scores, improving bonus allocation accuracy. Designed Tableau dashboards for workforce insights, increasing data-driven decisions by 35%. Mentored 2 junior engineers and collaborated with data scientists to deliver scalable pipelines under tight deadlines.