Sangeet Kumar

MAX-NG [APR 2023 - Present]

• Call Center Dashboard

Built reports for various KPI's and created leaderboard Dashboard where stakeholders can identify top and low performers based on all filtration metrics at each metric level

• Vehicle Data Ingestion

Created a cloud function to pick each file from a generic bucket and check the documentStatus and based on it move it to a specific GCS bucket and load it to a specific BQ table.

• Collection Agent Reporting Dashboard

Built Monetary Collection reports for each of the agents Collection performance for timeline intervals which helped to improve the agents collection analysis.

DrivenBrands [DEC 2021 - APR 2023]

• TEKMETRIC Royalty Issue
  Implemented ETL using DBT model from the source data and developed a DAG with External Task Sensor for multiple tables in task level in a parallel manner to undergo full refresh on a daily basis which helped the business team for better data analysis and recover the missing records.
• CHATTE R Integration For Multiple Brands

The source data available from each Brand is integrated into Business needed data by Streaming from the source, ETL process and populated it into the BQ table with Airflow through CI/CD Pipeline.

• SONNY Data Integration

The various sites in the SONNY'S CARWASH CONTROLS are pulled from the API endpoint and loaded into Temp Table. Created a DBT model for BQ tables for each category and ingested into the destination table and did DataOps work from the retrieved data and created a view.

• Google Cloud Composer Migration

Upgraded Apache Airflow 1.10.10 to Cloud Composer Airflow 2.2.5 version. I have Collaborated with the team on changing the configuration and DAG changes with respect to the Cloud Composer Environment.

• Google Ads API

Explored the Google Ads 360 API and gathered the information of generating the OAuth Token. Developed a DAG to ingest the missing data dump into the BQ table.

Built a DBT model and ingested in the Piperr, for data ingestion from postgres to BQ on daily basis with Airflow DAG developed from Piperr.

• Developed a Chargrid model which is used to preserve the 2D structure of the document and can be used to extract the meaningful information from the generic documents.
• In this project my contributions are Preprocessing, model Training and Prediction.
• Worked with On-site team for Discussing, reporting and presenting the weekly progress of the project.

Language, Frameworks and Tools

Python, TensorFlow, PyTorch, CNN, OpenCV, Cloudera Workbench for Data Science Cloud Platform, Docker, Jupyter Notebook, Kaggle, Google Colab, Git.