Preetham Kumar Madupuri

• Built Machine Learning models using state-of-the art ML algorithms like Deep Neural Nets, SVM, Logistic regression etc.
• Architected and implemented distributed data pipelines using Apache Spark for large-scale data processing, achieving improved scalability and reduced processing times for complex transformations.
• Developed real-time data streaming solutions leveraging Apache Kafka, enabling low-latency ingestion and processing of high-throughput event streams.
• Designed and deployed cloud-native data workflows on AWS using services like EMR, S3, Lambda, and RDS, enabling seamless integration and automated ETL processes.
• Implemented fault-tolerant and highly available data pipelines leveraging distributed processing frameworks like Spark and Kafka Streams, ensuring data consistency and durability across the system.
• Working experience with large scale machine learning environments in building and supporting which includes planning, designing, installing, configuring, performance tuning and monitoring.

• Built data pipelines using python to enable exploratory data analysis for NLP around speech-to-text from

Cloudera support calls.

• Significantly improved the performance of EMR jobs by 50% by implementing the latest developments around Spark SQL and EMR clusters

• Implemented hot-warm data architecture using AWS S3 and Apache Spark and this helped improve the

performance of call stats by ~70% and also reduced storage costs

• Ingested data from disparate data sources using a combination of Spark SQL, Google Analytics API, python to create data views to be used in BI tools like Tableau/Looker

• Designed and implemented batch and streaming pipelines with robust high availability to ensure product

uptime (delivered product uptime of ~99.999%)

• Developed and maintained complex ETL workflows in Java and Scala, processing billions of records daily to support business intelligence and machine learning initiatives.
• Integrated data pipelines with AWS services such as Redshift, Athena, and Glue, automating data processing and transformation for analytics and reporting needs.
• Managed Kafka clusters for stream processing and real-time data ingestion, ensuring optimal performance, partitioning strategies, and consumer group management.
• Improved system efficiency by tuning Spark jobs, partitioning strategies, and optimizing resource allocation, reducing compute costs by 20% while maintaining processing speed.
• Collaborated with cross-functional teams including data scientists, analysts, and product managers to deliver scalable and reliable data solutions tailored to business requirements
• Enhanced query performance by introducing partitioning & Bucketing in Hive.

• Some highlights are: - Automated ETL processes, making it easier to wrangle data and reducing time by as much as 40% - Increased the efficiency of the data fetching by using query optimization and indexing

• Constructed a data pipeline that helps detect entities from scientific articles and modified the entity extraction

process to help prepare training data and thereby helping in building classifiers using state-of-the-art Natural Language Processing techniques such as NER.

• Built a Django based web application tool

• Used Spark to implement scalable Machine Learning topic algorithms on large datasets.

• Improved company nomination system between various entities of interest that nominates links between companies, chemicals and harms

• Detected harms from boxed warnings using a combination of semantics and natural language techniques.