Sandeep Kumar

● Participated in IBM Watsonx Technical Enablement Bootcamp,
gaining hands-on experience with Watsonx tools for building and
deploying AI/ML solutions.
● Attended advanced training on Generative AI, focusing on
applications of large language models (LLMs), ethical AI
implementation, and real-world deployment strategies.
● Explored cutting-edge AI frameworks and solutions to enhance
expertise in designing scalable and impactful AI-driven products.

• Developed sophisticated models leveraging statistical analysis and machine learning techniques.
• Led projects focused on Logistic regression , Test Similarity, Summarization, and Sentiment Analysis.
• Reviewed business domains to bolster analytics, enabling informed decision-making backed by robust data.
• Gathered analytics data from diverse sources, ensuring comprehensive insights.
• Presented insights through compelling data visualizations, utilizing tools such as Tableau and Jupyter Notebook.

Project Title : Generative AI Research Initiative

Role: Lead and Contributor

Technologies: GCP Vertex AI , Python

The project was a generative AI initiative involving multiple teams

focused on developing various components, including a prompt

engineering engine and advanced chat capabilities .My primary

responsibility was to lead the implementation of the document

processing and retrieval system enhance user interaction with the

documentation.

As the lead, I oversaw the development and was involved in key tasks

such as:

- Extracting text from the PDF documents.

- Generating embeddings for the extracted text using the GCP Vertex AI model.

- Generating responses based on the stored PDF content using the GCP

Vertex AI

Project Title: App Review Analysis

Role: Lead and Contributor

Technologies: AWS , BERT, Python, Pandas

The project aimed to classify user sentiments from app reviews on the

App Store and Play Store, addressing the need for automated sentiment

analysis to inform product enhancements and user engagement

strategies.

-Led the development of a sentiment analysis solution to classify user

sentiments from App Store and Play Store reviews.

- Worked on AWS environment, prepared datasets.

notebook instance.

- Developed a sentiment analysis pipeline utilizing BERT and used

pandas for data preprocessing and analysis.

- Generated insights from user feedback, resulting in improved app

ratings and enhanced user satisfaction through targeted product

enhancements.

The project resulted in a structured dataset with sentiment labels and

scores, enabling data-driven insights into user feedback. This facilitated

targeted app improvements, enhanced user satisfaction, and improved

overall app ratings.

Project Title: Text Similarity Analysis
Role: Lead and Contributor
Technologies: AWS, Python, Pandas
Spearheaded Natural Language Processing initiatives, particularly in
Text Similarity Analysis.
Employed advanced tokenization methods to discern similarities
between textual descriptions.
Leveraged libraries including NLTK and scikit-learn to refine analysis.

Project Title: GCP Airflow Pipeline Monitoring System
Role: Lead and Contributor
Technologies: GCP Cloud Composer, GCP BigQuery ,Airflow, Python,
Jenkins
The project aimed to develop a system for monitoring various Airflow
pipeline jobs using the Airflow database. This initiative was essential for
ensuring the reliability and efficiency of ETL processes within the
organization on GCP environment
- Extracted data from the GCP Airflow database to monitor job
performance and status.
- Processed the data through an ETL workflow using Airflow.
- Saved the processed data back to Google cloud for analysis and used it
using BigQuery for reporting.
- Set up and managed Jenkins for continuous integration and
deployment of monitoring scripts.
The project provided real-time insights into Airflow pipeline jobs,
enabling better oversight and quicker resolution of issues. This
monitoring capability enhanced the reliability of data workflows,
contributing to improved operational efficiency across data processing
tasks.

Project Title: App Risk Score Exploration and classification
Role: Lead and Contributor
Technologies: Python, Pandas, AWS, AWS Boto3
The project aimed to classify app based on risk level
- Conducted thorough Exploratory Data Analysis (EDA) to extract
actionable insights.
- Applied advanced statistical analysis and logistic regression models to
classify apps based on risk levels.
- Correlation Analysis
- Conducted statistical analysis to unveil correlations between specific
operating system versions and associated risks.
- Utilized Tableau for dynamic visualization, enhancing comprehension
of complex relationships.

● Extracting data from AWS S3 and AWS Athena

● Writing code on python using GitHub as code repository and Gerrit as code review tool.

● Analysing and performing exploratory Data Analysis available in AWS cloud using python Pandas, Jupyter lab etc.

● Creating report on Tableau for few of the statistical analysis .

Production Triage report Analysis

· Led and worked along with the team on Data analysis & Visualization project where the dashboard was created for all the recent failure for the job running in airflow. This involved airflow database, creating airflow Dag , AWS S3 , Tableau.

· Extracting and processing data from AWS Athena and AWS S3

· Performing Statistical Data Analysis available in AWS cloud using python Pandas, Jupyter lab etc.,

● Coordinating with stakeholder/Business Owner for test management, Requirement clarification, UAT sign off.

● ETL Test Automation Through SQL and Python Pandas

● Creation of Master test plan, Test Reports, traceability matrix.

● Test case creation and Execution in JIRA.

● Data validation for data sourced from different systems like oracle, Wall Street etc. and different formats like XML, CSV, MQ etc.

● Writing SQL queries, working on Unix Environment, data stage environment.

● Testing banking Products like Calypso, Covered Bond, Wall Street etc.