Experience | Pradeep Senthil

Aug 2024 – Current

Refactored all ML Kafka message processing services (100K+ messages/day), cutting peak message lag by over 40% and reducing debugging time by half
Single-handedly migrated all ML codebases from AWS CodeCommit to GitHub and CI/CD pipelines to GitHub Actions, reducing deployment flakes by 90%+
Maintained Alto's data platform (Airflow, DBT, Snowflake, Looker) supporting 10+ analysts and 70+ business stakeholders
Upgraded production Kubeflow (1.6→1.9) with zero service interruption, reducing cloud infrastructure costs by 5x
Built production OCR framework using AWS Textract to extract structured data from prescription faxes, automating 2000+ prescriptions monthly
Collaborated with CTO, VPs and Directors to foster engineering excellence, upskilling data scientists on Kubernetes, Observability, and CI/CD

Jul 2023 – Aug 2024

Engineered bulk and on-demand ordering systems to decrease Out Of Stock (OOS) rate by 30%, improving inventory turnover and reducing costs
Reduced pricing system response time by 30% through data quality improvements and database schema optimizations, supporting 50k+ daily transactions
Eliminated 80%+ of inventory/pricing-related engineering escalations by building self-service automation tools, saving 10+ engineering hours weekly

Sep 2021 – May 2023

Designed and developed a deep learning analysis framework to process over 30 terabytes of swine research video into actionable metrics
Designed CNN algorithm to automatically segment brain gray matter from pig MRI images, achieving a mean IoU score of 94%. Orally presented at SBR 2022 research conference
Designed and developed AVAT, an open-source video annotation tool for behavioral and computer vision applications

Jan 2022 – Aug 2022

Developed an automated solution to generate a large synthetic deepfake dataset consisting of over 30,000 deepfake videos for internal analysis and training
Researched and developed POC deep learning models achieving 70% deepfake detection rate on the Facebook DFDC dataset
Created Doublecheck: Django models and management scripts that improved facial recognition system by reducing false positivity rate by over 80%

EDUCATION

Aug 2021 - May 2023

Masters of Science - Computer Science

Advisors: Dr. Matthew C. Caesar, Dr. Ryan N. Dilger

Courses: Deep Learning for Healthcare, Topics in Software Engineering, Topics in Internet of Things

Aug 2017 - May 2021

Bachelors of Science - Computer Science

Courses: Data Structures, UI/UX Design, Algorithms, Applied Machine Learning, Computer Security

Deep learning framework for automated MRI segmentation in veterinary research applications.

Research study on behavioral analysis in swine using computer vision and machine learning techniques.