Default profile banner
SP

Saurav Pandit

@sauravpandit

Data Scientist II at MasterCard India Services

Gurugram, Haryana

MasterCard India ServicesIndian Institute Of Technology Roorkee

Saurav Pandit is a Data Scientist II at MasterCard India Services with experience in building Kubeflow pipelines for malicious traffic detection. He specializes in deep learning, NLP, and unsupervised algorithms, utilizing technologies such as PyTorch, TensorFlow, and Spark. Saurav holds a B.Tech from IIT Roorkee and has a strong background in predictive modeling and data analysis.

Experience

Data Scientist II

MasterCard India Services

•Jul 2022 - Present•Gurugram, Haryana

Designed a two-step Kubeflow pipeline for near-real time detection of B2C and B2B malicious traffic. Optimised end-to-end processing and inference time by more than 30 percent by using both unsupervised and supervised learning for detecting malicious ips and calls. Applied unsupervised algorithms(VAE, RSRAE,DeepSVDD) on static and velocity api features,created using Spark in near-real time for highlighting ips and validated the user ip requests by supervised learning algorithms on requests traffic. Explored transformer based architectures(BERT) for generalising the problem of api-security. Refactored code from Tensorflow 1.x to Pytorch, parallelised the modules in production with minimal changes to recall and precision.

AI Intern

MasterCard India Services

Internship•May 2021 - Jul 2021•Gurugram, Haryana

Generate human-like tweets and hashtags on articles using GPT-2 and GPT-3. Implemented automatic clustering and extractive labelling of career-aspirations of MC employees. Researched on Bipartite GNNs for use in Mastercard environment.

Education

Indian Institute Of Technology Roorkee

B.Tech

Electronics and Communication

Aug 2018 - Jun 2022•Grade: 8.34/10.0

Doon Public School

Apr 2016 - May 2018

Skills

Data Analysis
Predictive Modelling
A/B Testing
NLP
Deep Learning
Blockchain
C++
Python
SQL
Solidity
Tensorflow
Pytorch
Spark
Impala
Scikit-learn
transformers
Huggingface
Matplotlib
Seaborn
AWS
Kubernetes
Jupyter
Flask
StreamLit
MLFlow
Jenkins
Docker
Kubeflow
Git
Github
Prometheus