Anoop Kumar

About Anoop Kumar

I am an experienced Data Scientist proficient in diverse machine learning algorithms, including advanced generative AI techniques. Skilled in Python, Machine learning, Deep learning, AI and feature engineering, with a track record of delivering impactful insights and innovative solutions. Passionate about pushing the boundaries of artificial intelligence through creativity. Excels in transforming data into actionable strategies to optimize processes and enhance user experiences. Eager to learn and adapt in collaborative environments. Let\'s connect.

Experience

1. Data Scientist

   Hexaview Technologies Inc.

   Aug 2023 — Present · Noida, IN

   As a Data Scientist, I\'ve worked on a Fraud Detection System tailored for US markets, aiding platforms in integrating our SDKs for document verification and user fraud detection. By analyzing user data, the system identifies fraudulent transactions, helping platforms verify the identity of customers during registration. It employs three distinct models to predict the likelihood of fraudulence in applicants.• Integrated various vendors to verify the identity details of the customer registering with the firm.• Introduced Pre-checks like KYC and Document verification for initial checking.• Divided the data into different types for various fraud types that could be encountered.• Applied Feature Engineering by creating various features from different vendor calls and applying them to preprocess, building various new and average features for the model.• Analyzing all the features and the impact on the decision using Shap plot and correlation scores and selecting the top preferred features for training.• Trained various full end-to-end fraud detection models using H2O.ai on different selections of features and compared their performance to select the best• Applied optimization and improvements on the selected model while ensuring that overfitting is avoided.• Trained 3 different models for each fraud type and generated an aggregate decision on the user.• Productionizing the Fraud Detection Models for application use.• End-to-end testing and individual feature checks from vendor responses for each transaction. • Continuous monitoring of the performance of the model by performing backtesting to check the improvements being made over time and analyzing the result.

Education

• Meerut Institute of Engineering and Technology(MIET)

  Bachelor of Technology - BTech

  2019 — 2023