Akash Raj Email & Phone Number

Seattle, Washington, Us

Santa Monica, California, Us

Developed a robust forecasting model with low error to predict the weekly sales of products using LGBM and XGBoost models.Implemented a product clustering framework using KMeans DTW to identify groups of products with similar lifecycle patterns.Utilized cosine similarity, sentence embedding and Levenshtein distance algorithms to calculate the similarity between the album names and product descriptions with around 70% accuracy, enabling efficient identification of artist and album names.Analyzed a comprehensive set of features to design a PoC to predict the demand for new vinyl products before the release.

Boston, Ma, Us

Coursework - Time series forecasting, Simulation (Monte Carlo), Optimization

Dubai, Dubai , Ae

Developed data Models and Explores in Looker using LookML scripts to automate calculation ofcomplex supply chain metrics which saved 26 person-hours per week. Also, developed a self-serving dashboard in Looker and PowerBI to track supply chain KPIs and generate actionable insights.Designed a framework to estimate the dollar impact of out of stock items and democratized the insights using an RShiny dashboard.

Bangalore, Karnataka, In

Customer Centric Product Assortment - Predicted sales for the future season of apparel products to optimize the store inventory, which led to an annual savings of ~USD 2M, by developing ensemble regression model of Random Forest and Gradient Boosting. Also, developed an R Shiny dashboard to share recommendations with buyers and insights to the stakeholders.Demand Forecasting - Streamlined the OTB plan (purchasing budget) of the brands and reduced over/under-buy by ~8% compared to previous season, by building a time series model, using ARIMAX, ETS and Prophet, to forecast monthly category sales.Markdown Optimization - Created a framework to dynamically markdown product MRP over the season, based on its performance, which increased the sales by 4% in the test stores as compared to the control stores, by implementing Machine Learning and Price Elasticity models.

Northbrook, Il, Us

Patient Risk Prediction and Outcome Profile Generation – Collaborated with healthcare professionals to identify patients with high risk and gain better insights throughout the patient engagement journey by building a classification model to predict patient risk profiles obtained from patient electronic medical record data using Logistic Regression with an accuracy of 92%.Custom Analytical Workflow for Post Market Surveillance – Led a 3-member team to develop a self-serving, centralized repository of R package to set up a post-market surveillance framework using complaints data, employing statistical process control tools like Shewhart, EWMA, Poisson Western Electric signal detection.

Northbrook, Il, Us

Brand Adopter Lift Analysis – Developed and automated different quantitative approaches to measure incremental sales lift of brand adopter households (test) to the control households using Propensity Score modelling and Bayesian Time Series modelling.Store Prediction for expansion initiative - Matched the adopter households to the non-adopter households based on propensity score, calculated using transaction, demographics and CLTV household metrics, to analyze the sales lift of the adopter over the non-adopter. Developed a classification model to classify stores where the initiative is more likely to be a success using Logistic Regression and Random Forest, and the most significant variables were selected out of a pool of candidates as the actionable levers. This assisted the retailer launch the initiative in specific stores which accounted for an incremental sale of ~$2.4mnReproducible Framework for Tracking Travel Expenses - Developed automated the ETL processes of data from multiple sources (CSV in Box.com) to consolidated tables in Hadoop. Built a custom application to integrate Box.com with Hadoop to execute ETL process. Used Python and Cron to automate and schedule the report creation process, minimizing effort from 19 person days to 3 hours.

Master'S Degree, Analytics & Machine Intelligence

Bachelor Of Technology - Btech, Information Technology