SUSHMA KARJOL

+*- (***) -***-**** adwrmh@r.postjobfree.com linkedin.com/in/sushma-karjol github.com/sushmakarjol EDUCATION

NORTHEASTERN UNIVERSITY Boston, MA

Masters in Analytics (GPA 3.85/4) April 2023

Related Courses: Analytics System Technology, Predictive Analytics, AI, Data Mining Application and Big Data. TECHNICAL SKILLS

Tools : SQL Workbench, Visual Studio code, GitHub, Microsoft Office, Jupyter Notebook, Anaconda, Apache Spark Hadoop, MapReduce, Hive

Languages : Python (Scikit-Learn, Numpy, Pandas, SciPy, OpenCV, Matplotlib, Geopandas, Seaborn, Tensorflow), R (dplyr, tidy verse, Ggplot2, Caret, Plotly, funModeling), SAS, Java, C/C++, SQL, HTML/CSS, Scala Database/ETL : MySQL, NoSQL, Postgres, SQLServer, Snowflake, Hadoop (Hive, HQL, HBase, MR), PySpark, OracleDB Talend, Kafka

Analytics/Reporting : Tableau Server, Tableau Public, Power BI, Qlik, MS Excel Solver, Alteryx, Google Data Studio. Cloud Experience: Amazon Web Services – S3, Lambda, EC2, Apache Kafka. ML Models : Supervised Learning (Ensembles, Bagging, Naive Bayes, SVM, Deep learning), Unsupervised (K-nearest Neighbors, K- means clustering, Neural Networks, PCA, Anomaly detection), Reinforcement Learning Markov Decision )

RELEVANT PROJECTS. BOSTON,MA

AI Application & Big Data Sep 2022 - April 2023

MARKET BASKET ANALYSIS( SPONSOR - URBAN VALUE STORES) Power BI, Python

• Cleaned the unstructured data, applied RFM (Recency, Frequency and Monetary) techniques to identify the product groups and customer segmentation to the basket required for cross-selling. AI - NLP(Text mining), Keras framework(Vision AI), A-star Algorithm(Robotics) Google Colab

• Applied NLP Pipelines to clean Unstructured tweet messages to analyze public sentiments and developed a Convolution Neural Network model. Recreated the classification CNN model to classify text using hyper-parameter methods and revised model accuracy to 87.65% with 5 iterations of Cross-Validation.

• Programmed path using A* Algorithms. Applied image processing techniques using OpenCV computer vision library. WORK EXPERIENCE

RUE-GILT GROUPE Boston, MA

Business Intelligence and Data Warehousing Engineer Jan 2022 - Jun 2022

• Presented interactive dashboard, used complex SQL queries,ascertained decision to reduce operational cost of around 100K per year, by tracking outdated/slow-moving products and replacing with high-demand products by observing Trend lines.

• Built performance Analytics dashboards and helped an internal Engineering Team enhance the performance of upcoming internal tools by saving up to 2-3 hours per day of troubleshooting. (Or 10 -15 hrs. of processing time per week). RELEVANT PROJECTS. BOSTON,MA

DATA VISULIZATION ANALYTICS, ML & STATISTICAL ANALYSIS. Jan 2021 - Dec 2021 CUSTOMER SEGMENTATION - (SPONSOR DATA CLUB)Python Jupyter-notebook.

• Delivered customer insights and formulated Data mining techniques -Clustering analysis (K-mode), Gradient-Boosting Classification and Logistic Regression on the Banking data with 89-90% model accuracy. FINANCIAL ANALYSIS - (SPONSOR ALICORN VENTURE CAPITAL)PCA:Python & Qlik

• Predicted Stock performance using - The stock market in the United States assessed 20% greater returns than other regions, with stock series A' and B' ranging between 18-20% margin returns under cloud technology businesses in the IT industry. SUPPLY CHAIN- Demand Analysis (NEU) R Markdown

• Built General Linear Regression Model, regularized coefficients using Ridge and Lasso techniques, identified 17% of product-demand as falsely categorized by applying Confusion Matrix on regression analysis. CONSUMER BEHAVIOR – Hypothesis Testing (NEU) R Markdown

• Designed interactive data visualizations in Qlik and offered significant business insights and analysis based on comprehension of visualization. Studied consumer behavior by conducting Hypothesis Testing and Chi Square test. HOUSING EVALUATION – Supervised Classification Models(NEU) Python Jupyter-notebook

• Programmed Classification models to predict if the houses were overvalued or under-valued. Random Forest and Gradient Booster had the highest classification accuracy with each 72%.