Utkarsh Bajaj

•DataMind: Architected and led the full-stack development of an AWS Bedrock-based GenAI RAG platform, resulting in 51+ monthly active users and 1000+ conversations, projecting 600+ engineering hours saved annually. •Finny (Agentic Forecasting Engine): Established and launched an LLM-powered forecast fine-tuning recommendation engine via a Streamlit web application, reducing manual analysis time by 70% (2,400 annual hours saved).

•SOP Sherlock (Semantic Search): Designed three-LLM architecture for semantic content similarity detection beyond keyword matching, earning featured poster recognition at Amazon Analyticon conference. •Unified Data Model (UDM): Designed scalable PostgreSQL data model with governance framework (RACI, SOPs) for 11 functions, including validation scripts reducing discrepancies by 80% for onboarded functions. •MLPigeon (MLOps Standardization): Resolved package dependency issues and enhanced SageMaker platform for ML standardization, enabling smoother onboarding and reduced maintenance.

•Developed an automated ensemble forecasting model of investigation task volumes to hire manual investigators for Amazon. Improved forecasting accuracy from 11% in 2022 to 8% in 2023 by statistical analysis and driver identification. •Automated the forecasting process using python scripting enabling the team to generate forecasts within 10 minutes as compared to 5 days. •Created ETL data pipelines and AWS QuickSight dashboards for 5 business functions across the 3 team verticals to allow leaders to get a uniform view of forecast variance. •Devised an app-based headcount optimization solution on Python using Pulp optimizer enabling workforce planning teams to generate optimal headcount instantly.

•Established a mixed linear integer optimization model across 5450 retail stores in United States utilizing cvxpy package in Python maximizing the profit potential of each store by 11.24% •Performed SKU assortment as a part of prescriptive analysis by assessing shelf space and identifying gaps in supply chain therefore maximizing category and store coverage by 97% on average across all stores •2nd runner up at the 2022 INFORMS Business analytics Conference - Sponsored by SAS (Houston, Texas) •Best paper award at 2022 MWDSI Conference

•Led a cross-functional team of 4 members to monitor audit compliances by integrating MySQL and Oracle ERP data monthly thereby produced valuable insights; helped increase the compliance rate from 65% to 85% in Quarter 1 of 2021 •Ensured seamless procurement of these items during covid imposed lockdown; modeled price by ARIMA forecasting saved 6 million USD of CAPEX in that quarter •Developed a metric-based mechanism of ranking supplier performance in VBA and visualized it by QlikView; improved monthly on time delivery by 23% and decreased monthly defect rate by 1.5% for major suppliers

•Predicted the likelihood of customer churn using logistic regression in the Direct-To-Home business leading to a 2.3% decrease in churn and 26% increase in profitability •Deployed a contract management tool on Excel for quantifying commercial clauses of contracts for all goods and services across Direct-To-Home division, further digitized 92% contracts on central contract repository tool Icertis •Spearheaded audit of processes of third-party logistic providers; validated and optimized the processes saving 3.3 million USD in a single fiscal year

•Devised 50+ robust alphas by converting mathematical expressions into Python code to predict performance of financial instruments; achieved returns more than 10% above average •Recommended major trading strategies in US and Asian equity markets by developing algorithmic procedures

•Understood wireless communication systems and different network architectures for 1G,2G,3G,4G, and 5G technology. •Tracked and monitored all National Long distance (NLD) and Point of Interconnect (POI) calls. •Developed a mechanism to report the number of lost calls across India and implemented strategies to reduce them.

•Understood intricacies of functioning of a Stock Exchanges and a Broking company. •Assisted in trading, settlement of transactions, risk management, client acquisition, and registration.