Prasoon Kumar

Company Overview: Epsilon, Publicis Groupe Company

Responsible for :

• Led the strategic design and development of scalable data engineering solutions across Databricks and Greenplum, integrating Notebooks and workflows for ETL orchestration, implementing CI/CD pipelines using Databricks Asset Bundles, and establishing robust data lineage, governance, and security through Unity Catalog.
• Demonstrated deep technical expertise in ETL, SQL, PySpark, and modern data engineering practices, while serving as the primary point of contact for all data-related initiatives—including architecture design, solution development, and operational excellence.
• Architected and managed high-performance ETL/ELT pipelines across Databricks and traditional databases, integrating diverse data sources to power business intelligence and operational reporting.
• Built scalable data infrastructure, including Data Lakes, Lakehouses, and Data Warehouses, ensuring optimized data management, performance, and broad accessibility across the organization.
• Enabled the team to deliver robust and scalable data pipelines by designing and deploying Apache Airflow DAGs, significantly improving the efficiency and reliability of ETL workflow orchestration.
• Architected and developed an analytics platform (CIP) on AWS, enabling insights into consumer sentiment across social media and search engines using data from Netbase and Google AdWords.
• Ingested and processed large-scale social and web content, applying machine learning models in AWS SageMaker to accurately group user-generated terms with spelling variations, paraphrasing, and rewording, using supervised and unsupervised learning techniques.
• Applied K-means clustering and PCA for dimensionality reduction and high-variance identification, followed by cosine similarity scoring for cluster merging and normalized volume ratio calculation.
• Performed automated cluster labeling using Gensim LDA and Markov Chains, then trained a labeled dataset on a Deep Learning Ensemble Model (Normalized CNN + BiGRU).
• Utilized Docker images via AWS ECS to package ML models, generate artifacts, and deploy real-time endpoints.
• Provided strategic leadership and mentorship to a high-performing team of data engineers, aligning their initiatives with organizational objectives and driving measurable business outcomes through innovative, data-driven solutions.

• Company Overview: GCC Bangalore

Responsible for :

• Building and enhancing EDH Capability by Architecting, Designing, developing reusable ETL frameworks In Talend.
• Leading and governing Finance, People and Supply Chain streams end to end to ensure high quality deliveries and provide technical mentorship with driving process improvements, standardization and standard methodologies.
• Designing Audit Balance Control framework to orchestrate entire ETL pipeline.
• Designing and creating ETL Pipelines to ingest data in Data Lake using Hive (using spark execution engine) and ADLS, used ORC file format and PySpark.
• Creating ETL pipeline and using Python module to transform huge amount of data.
• Also created translation framework in python using pandas, google trans etc.
• Designed Hive tables to store very huge history data, which is consumed by data scientist for exploration, analytics ad-hoc queries.
• Defining Data Quality rules, Data lineage, data catalog for multiple systems.
• Designing Relational and Dimensional data model in Erwin by integrating multiple system like Sharp, Navigate, Click etc.
• Identify Critical Data Elements (CDE’s) from Enterprise Data Model.

• Company Overview: GE Digital

• Company Overview: Markem-Image(Machinary)