Summary
Overview
Work History
Education
Skills
Certification
Accomplishments
Timeline
Generic

GUDURU SAI DEEPTHI

Hyderabad

Summary

Adaptable Data Scientist with extensive experience providing first-class results. Meets job demands and deadlines through diligent work-ethic and dedication to quality.

Overview

1
1
Certification

Work History

Associate Data Science

National payments corporation of India
Hyderabad
08.2023
  • Created various autoencoders tailored to specific fraud scenarios
  • Implemented a Slabbing algorithm to categorize numerical features by the skewness of their data distribution, thereby preventing model overfitting
  • Created innovative solution using katz algorithm to identify suspicious nodes in UPI and IMPS transactions, allowing for analysis of money movement patterns through multiple hops
  • Developed and implemented a Graph neural network to analyze money transfer patterns in potentially fraudulent entities using data obtained from aforementioned method
  • Developed a Recursive Batch clustering algorithm that efficiently clusters large-scale data (60cr
  • 88 Features), categorizing UPI profiles into different fraud and non-fraud categories through recursive batch processing
  • Incorporated the use of Cosine Similarity for dimensionality reduction, effectively transforming 88 features into a compact and manageable 3-dimensional subspace
  • Applied Positive Unlabelled (PU) learning to identify similar suspicious behaviors within a vast unlabeled dataset using a small labeled fraud sample
  • Utilized Dagster to create and organize SQL pipelines for generating daily profiles, facilitating real-time model inferencing
  • Managed the development of kubeflow pipelines converting generated profiles into Redis command file format and pushing them to Redis for model inferencing
  • Developed various dashboards in Superset using Druid and Trino sql engines for real-time monitoring of the model and to publish model performance metrics to higher management.

Graduate Engineer Trainee

Hyderabad
10.2022
  • Oversaw the development and maintenance of UPI transaction data porting pipelines, facilitating the transfer of information from numerous databases to the Hadoop file system employing Airflow
  • Implemented efficient data flow from kafka to hadoop using Spark to ensure seamless model log retrieval
  • Conducted exploratory data analysis on reported fraud data in UPI platform utilizing SQL and Python
  • Engineered key features such as birth date and new interactions to effectively differentiate between fraudulent and nonfraudulent activities
  • Developed lightweight Random Forest model to detect and prevent fraudulent transactions in real time by exploring multiple use cases
  • Utilized Hive query language to extract UPI data, facilitating the development of informative Tableau reports used for strategic business decision-making.

Education

Bachelor of Technology - Computer Science Engineering

Malla reddy college of engineering
01.2018 - 1 2022

Skills

Python

Certification

PG Certification in AI-ML, Internation Institute of Information Technology, 2022, 2023

Accomplishments

  • Earned "NINJA" badge in the organization's coding contest in MAY 2024.
  • Received "WELL DONE" award for my contribution in building fraud models and reducing frauds.

Timeline

Associate Data Science

National payments corporation of India
08.2023

Graduate Engineer Trainee

10.2022

Bachelor of Technology - Computer Science Engineering

Malla reddy college of engineering
01.2018 - 1 2022
GUDURU SAI DEEPTHI