Tony Hammack Email & Phone Number

Birmingham, Al, Us

- Leading the OCR initiative by using deep learning and computer vision to enrich Shipt's catalog and improve efficiency (DocTR & DONUT for analyzing receipts; Tesseract, DocTR for product images)- Lead the Cost Correction initiative by using optimization techniques to improve prices for shoppers and increase Shipt's profits - Lead the Scrape Optimization initiative by using integer programming to find correlations between stores and their product prices. This allows for considerable savings and is an important feature for other projects- One of the lead maintainers for Shipt's Internal ML Tooling

Birmingham, Al, Us

- Improved cost accuracy, ensured regulatory compliance, and reduced fraud on Shipt’s platform by applying OCR, computer vision, and deep learning techniques to uploaded receipts (Python, Pytorch, OpenCV, CNNs, Transformers, Snowflake, Airflow, Streamlit)- Implemented an image hashing strategy for Shipt’s product images - used as feature in downstream pipelines (Python, Pytorch, PaddlePaddle, OpenCV, Snowflake, Airflow)- Created a NLP pipeline which matched extracted OCR data from receipts to ordermetadata (Python, Rust, Pytorch, Transformers)- Worked on Shipt’s internal ML platform (Python, DroneCI, Snowflake, Airflow, CometML, MLFlow, Redis)- Worked on Shipt’s new feature store implementation (Python, FastAPI, GCP BigTable, Snowflake)- Contributed/maintained various Python support packages used by Shipt’s data scientists and machine learning engineers (Python, Feast, Snowflake, CometML, MLFlow, Onnx)

Birmingham, Al, Us

- Develop and integrate computer vision models (Python/Pytorch/C++)- Use Kotlin and AWS KDS/Kafka to deploy computer vision models for near-real timeinference- Develop Javascript SDK (browser and Electron) for WebRTC streaming- Develop API for video session management- Support current video infrastructure (WebRTC, Websocket, and Spring Boot)

- Develop data labelling and experiment tracking platform Lit- Create data extraction tool for bed exit project- Develop model architecture (from 55 percent to 98 percent accuracy) for bed exit project - Use classic computer vision techniques for counterfeit detection project- Worked on various model architectures for EEG diagnosis project

Birmingham, Al, Us

Master Of Science - Ms, Mathematics

Bachelor Of Science - Bs, Mathematics