Sandeep Athota

About

AI/ML engineer at J.P. Morgan Chase who deployed a production financial-risk prediction platform combining CNN/LSTM/gradient boosting on AWS SageMaker, with automated drift-triggered retraining and governance-grade fairness testing. Leveraged SageMaker Clarify plus SMOTE and LLM-generated synthetic data to improve minority-group F1 by 0.12, and communicated results to non-technical risk/ops teams via Power BI dashboards.

Experience

Education

Key Strengths

• Built and deployed multi-model financial risk prediction system in production (CNN/LSTM/GBM on AWS SageMaker)
• Scaled ML over ~950k records and ~1.4M documents for risk pattern detection
• Automated drift-triggered retraining with AWS Lambda + SageMaker Pipelines and threshold-based validation
• Implemented fairness/bias evaluation across demographic and account segments with audit logging in MLflow
• Improved minority-group performance/fairness (minority-class F1 +0.12) using SMOTE, class weighting, SageMaker Clarify, and LLM-generated synthetic data
• Maintained production performance (reported >94% F1) with monitoring via MLflow and SageMaker Model Monitor
• Strong orchestration experience: Kubeflow for ML lifecycle pipelines; Airflow for ETL scheduling and retraining triggers
• Effective cross-functional communication: translated model outputs into Power BI dashboards for risk/operations stakeholders and iterated on KPIs/thresholds