• Enhanced service-levels of 4 distribution facilities (catering to 30 stores) by optimizing safety stocks, on-shelf-availability, warehousing, and ordering costs, preventing stock-out losses worth $ 200K for May 2024 • Setup PySpark ETL pipeline to fetch and apply transformations on SAP HANA real-time data stream while carrying out 6 statistical tests (F-test, Chi-Square, ADF, ACF, PACF, Granger causality) using Spark MLlib and RDDs• Forecasted safety stocks for top 18 revenue driving stock-out vulnerable product-vendor groups, by building an ensemble of: Bayesian model (demand uncertainty), Poisson model (short-term sales peaks), and Facebook Prophet (baseline)• Applied SAS Viya to provide insights, study effects of interaction terms (weather, fuel price, convenience) on model forecasts

THiNC.technologies subsidiary• Integrated GPT 2 transformer into PySpark cluster for user profiling, and analyzing customer queries and purchase patterns• Monitored chat-bot via tableau dashboards, demonstrating 76% utterance-intent-context match, improving overall digitization• Reduced database (30+ entities) query response time from 2 min to 15 sec, by executing PL/SQL triggers and cursors

• Collaborated at Imaging and Deep learning lab, team of 20+ researchers, dealing with disease detection and structural analysis• Migrated Raman spectral simulation scripts from MATLAB to Python, implementing clustering (GMM) on 1133 compounds• Trained a CNN-based autoencoder (Keras) to de-noise spectral images, leading to an F1-score of 87%

• Generated product recommendations for e-commerce app using collaborative filtering, associate rule mining, and customer review sentiment analysis, achieving a scale of 5000+ products, 80+ categories, and increased revenue by $ 60K (4.5M INR)• Automated ML pipeline, feeding data into supply chain dashboards through AWS SageMaker, Step function, Lambda triggers• Developed and maintained merchant app (admin) with HTML, CSS, Js, Django, MongoDB, NGinX, Docker, and AWS EC 2• Facilitated communications among stakeholders, sales team, and engineering team using MS Office (Excel, PPT, Word) and Jira