| | | 20,000+ Fresh Resumes Monthly | |
|
|
| |
Click here or scroll down to respond to this candidateCandidate's Name
Tel: PHONE NUMBER AVAILABLE EMAIL AVAILABLE LINKEDIN LINK AVAILABLE github.com/pr-Street Address EDUCATIONDuke University, Durham, NC May Street Address
Master of Science (Data Science) Relevant Courses: Computer Vision, Deep Learning, Machine Learning, Causal Inference, Natural Language Processing, Data Engineering, Deep Reinforcement Learning, Biostatistics, Statistical Modeling, Practical Data Science Tools, Data Visualization Indian Institute of Technology, Roorkee, India May Street Address Master of Technology (Power System Engineering) Relevant Courses: MATLAB, Computer Aided Power System Analysis, Machine Learning (Thesis) Uttar Pradesh Technical University, Lucknow India May Street Address Bachelor of Technology (Electrical Engineering) Relevant Courses: Linear Algebra, Probability and Statistics, Computer Programming in C, Discrete Mathematics TECHNICAL SKILLS Languages: Python, R, SQL, Bash, C, C++, MATLAB Database, Big Data & Visualization: Pandas, GeoPandas, Dask, MySQL, PostgreSQL, Spark, Hadoop, Matplotlib, Seaborn, Tableau, Power BI Software Tools & Cloud: AWS, Azure, Databricks, GCP, BigQuery, Dask, Docker, Hugging Face, Snowflake, Git, Airflow, Kafka, RedShift Machine Learning Framework: NumPy, PyTorch, TensorFlow, NLTK, Scikit-learn, SpaCy, Keras, Hugging Face, Langchain WORK EXPERIENCEEnergy Data Analytics Lab, Duke University, Research Assistant August 2023 January 2024 Instrumental in developing a high-resolution machine learning model using satellite-derived features to estimate global greenhouse gas emissions. Streamlined 600M+ rows of geospatial data processing by leveraging python scripting and multi-processing techniques using Dask, GeoPandas, geocube, and GDAL, resulting in optimized workflow for estimation of global CO2 emissions at a 1-km resolution. SunFi, Data Science Intern May 2023 August 2023 Improved the detection and prediction of customers' Probability of Default (PD) by ~40% by deploying an end-to-end credit risk model pipeline on cloud, collaborating effectively with cross-functional teams, using a range of cost sensitive predictive machine learning models. Implemented Synthetic Minority Oversampling (SMOTE) and feature engineering, for optimization of feature significance in credit risk analysis, resulting in a 20% boost in model performance and more reliable risk assessments. Analyzed Shapley Additive explanations (SHAP) values for interpretable clustering analysis, to gain insights into shared financial behavior. Identified anomalous transactions and potential fraud by implementing outlier detection methods like Isolation Forest. Enhanced preprocessing by fine-tuning a pre-trained NER model on bank statements to extract key entities, thereby improving the accuracy of fraud detection. KK Footwear, Data Scientist November 2017 March 2022 Streamlined operations by implementing inventory forecasting using statistical techniques (ARIMA, ETS) and machine learning models (e.g., LightGBM) capturing seasonality, trends, and sales variability. Reduced stock-outs by 20% and quadrupled annual inventory turnover. Expanded brand portfolio (1 to 5 brands) and retail partnerships (19 to 124) by implementing ML models for market trend analysis and product recommendations, leveraging sentiment analysis, topic modeling, and NER on customer reviews and social media data. Analyzed big data of 300M+ rows from 5+ sources to optimize sales strategy using statistical methods like regression and clustering techniques and created interactive dashboards for stakeholders in Tableau for real-time monitoring of the sales performance. Designed and implemented 10+ ETL pipelines for data warehousing to operationalize data for improved data quality and extraction of KPIs from sales data. This data-driven resulted in a 30% increase in Average Order Value (AOV) and an additional INR 13.5 million in revenue. Larsen & Toubro-Sargent & Lundy Limited, Power System Engineer August 2016 July 2017 Engineered comprehensive load flow and system reliability studies, for consulting projects, integrating machine learning for predictive short circuit analysis and optimization of transient stability, enhancing power plant performance metrics for outage and routine management. RESEARCH & PROJECTSAssessing Look-ahead bias in LLMs January 2024 Present Evaluate the extent of familiarity bias and look-ahead bias in LLMs when analyzing earnings call transcripts for downstream tasks. Develop mitigation strategies using a combination of company anonymization and prompt engineering. Debiasing Machine Learning models for Financial Services August 2023 May 2023 Collaborated with 2OS, to improve fairness, using techniques like matching to control for confounding variables and algorithms like reweighing, in a Gradient Boosting (XGBoost) model, to improve the Disparate Impact across race and gender by 40% without substantial loss in accuracy. Causal Inference Analysis on COVID Policies Feb 2023 April 2023 Conducted causal inference study on COVID policies, employing propensity score matching, difference-in-difference, and instrumental variable methods, identifying 52.37% reduction in new cases, highlighting the efficacy of policy interventions. Sentiment Analysis Web Microservice with AWS November 2022 December 2022 Developed a sentiment analysis microservice with FastAPI and Swagger, leveraging AWS S3, Lambda, and Comprehend for scalable analysis. Implemented CI/CD workflows with GitHub Actions, CodeBuild, and Docker for efficient build, test, and deployment. |
