BABLOO KUMAR

Developed hybrid search pipeline for ranking of research documents at Gartner.com, including creating index/query pipelines in Solr9 for BM25 keyword based search and vector index/query pipeline. Final re-ranking was done using cross-encoder trained on clicked data. Designed a real-time query vector pipeline. Trained gartBert embedding model using transformer Bert model api on Gartner documents data.

Designed algorithms for blog-based job and course recommendation systems and developed real-time personalized push/email notification system. Designed job recommendation re-ranking system using Listwise ranking algorithms. Analyzed user event data to prevent churn. Architected and implemented job domain and zone classification model. Created a tool to summarize job information pages. Worked on location, education, and job title standardization. Designed skills extraction tool for job and resume data.

Architected and implemented a deep learning based real-time personalized notification system for Economics Times, achieving ~2X growth in CTR. Led recommendation engine initiatives involving the design, development, and deployment of algorithms impacting ET core products. Developed and implemented Session based deep news recommendation engine, Collaborative filtering, and Keyword & Section based Recommendation Engine. Used NLP techniques for Keyword Extraction and word2vec, Support Vector Machine & Topic Modelling for Section categorization. Developed a Keywords Extraction tool using NLTK library in Python and Flask package for API purpose.

Developed a model to assign optimum number of dumper, shovel, and loader for minimum operating cost (Allocation Planning Strategy). Developed a transportation algorithm based model for dynamic dispatching of dumper in real-time. Developed a model to predict vehicle component failure using feature engineering and multi-class logistic regression methods, reducing delay time and increasing productivity.

Developed a statistical model for risk-based site monitoring using key-risk indicators. Implemented process automation including data collection, data processing, and statistical model development. Developed a model using various machine learning algorithms (SVM, Random Forest, KNN) to classify clinical trials issues and predict the duration of closing time.

M.Tech Thesis: Effect of Degree Of Financial Openness On Stocks Time Varying Betas. Estimated time varying beta’s of BSE-100 stocks using Kalman Filter Algorithm and developed R code for time series analysis (ARIMA, GARCH).