Devashree Buch

Validated the cleaning process by comparing pre and post-cleaning data volume and checking for data anomalies, with a targeted error rate below 1%. Constructed a Machine Learning pipeline with three stages, including a regression stage to build a model that predicts the SoundLevel based on other columns in the dataset.

Achieved 95%+ data accuracy when identifying trends or performance issues within financial statements. Analyzed a minimum of 3 to 5 investment portfolios monthly, identifying at least 2-3 key risks and providing actionable recommendations based on current market trends.

Automated ETL pipelines using PySpark in Databricks, achieving a reduction of over 20 hours of manual data processing weekly. Designed Tableau and Quicksight dashboards using SQL to connect Snowflake, Hive, and ACBS providing real-time financial insights to leadership & cross-functional teams.

Architected and constructed the data warehouse infrastructure and schema on AWS Redshift leading to a ~25% reduction in time spent on data integration and monitoring processes.

Analysed and generated 50+ detailed reports using Salesforce reporting tools to extract actionable insights from marketing data; drove a 40% increase in operational efficiency.

Drove an ML project to create deep learning models to enhance threat detection accuracy & reduce false positives by 30%. Trained and tested machine learning models using AWS SageMaker and EC2 for scalable data processing and model deployment.

Improved accuracy by 13% from 73% to 86% using L2 regularization.

Created innovative software solutions using Java and Hibernate framework, resulting in a reduction of data processing time by 25%.