· Designed and deployed scalable multi-strategy Gen AI training pipelines for Pre-training and SFT for text-to-video models.

· Integrating semantic, aesthetic, and composite quality filters to curate 100M+ high-resolution video samples of data—boosting data diversity and downstream model performance.

· Tested and benchmarked model and video decoders (pytorch) for checking performance, accuracy, measuring memory efficiency/usage and frame quality to ensure reliable data processing for model training.

· Developing, evaluating and optimize on large scale Machine Learning systems on Sagemaker pipeline (using Spark).

· Design, implement, automate and maintain scalable data system includes ETL/ELT pipeline, and deploy it.

· Design and implement data model, pipeline that include event driven and batch processing (ETL/ELT), data-warehousing, ML system pipeline, micro-service, and optimization on AWS. (DynamoDB, lambda, Databricks, DAG, step-function, Sagemaker, etc.)

· Lead and developed API service for full-stack web-application (Internal Analytic tool + Using foundational LLM model), which used react.js, node.js, AWS (DynamoDB, lambda, API gateway, ECS), docker, EKS(Kubernetes) and REST API.

· Lead multiple projects and manage and guide junior/mid-level engineers and build partnership with cross functional teams.

· Developed scalable systems leveraging supervised/unsupervised models, specifically anomaly detection Model optimization.

· Developed, tested, deployed and maintained E2E data pipeline for both batch and near-real time process.

· Managed and test pipelines, deployment, and monitoring on AWS: S3, Step-Function, Lambda, Sagemaker(productionalize), EMR.

· Mining, segmenting and analyzing in a variety of areas in Big Data Environments include Cloud data warehouse solutions (OLAP).

· Creating Automated ML and test Pipeline includes Data management pipeline (ETL/ELT) with web crawling, API, Python, Airflow and SQL. Source information from various external sources, such as Google, Youtube, Wikipedia, and Rentrak.

· Building predictive model for evaluating contents using Machine Learning techniques include NLP and Time-Series Analysis.