Chirag Shetty Email & Phone Number

Seattle, Wa, Us

• Led analytics and science initiatives for Amazon’s last mile driver staffing boosting operational efficiency by 30% within 6 months • Developed hiring projections models using k-means clustering to segment workforce patterns, LSTM time series forecasting for demand forecasting, reducing overstaffing with a $60M bottom-line impact• Centralized ML pipeline using Docker, CI/CD automation for model retraining & versioning, reducing deployment time by 80%• Achieved $21M marketing budget impact by developing a distributed media mix model using a log-linear approach to capture spend saturation dynamics and Bayesian optimization for real-time adjustments for maximized ROI.• Led data integration across marketing vendors, enabling real-time CPA cost monitoring with Apache Kafka for real-time streaming and Pyspark for processing petabyte-scale datasets, reducing response time, cutting campaign costs by 15%

Seattle, Wa, Us

• Developed heuristic approaches & BERT language models on customer-agent transcripts to generate coaching recommendations based on usage/feedback to improve agent performance, resulted in $20M annualized savings.• Applied active learning framework to reduce uncertainty on BERT model predictions, reduced annotation costs by 70%• Formalized strategy to develop annotation pipeline for training data and developed ‘Kudos’ recommendation model to reinforce positive customer interactions based on data from customer feedback surveys and contact transcripts• Led A/B testing to optimize recommendation ranking through gradient boosting methods, improved adoption by 12% • Leveraged sentence transformers (SBERT) to improve cosine similarity for help search results, increased accuracy by 18%

Seattle, Wa, Us

• Drove the vision for a notifications system for Amazon’s customer service agent performance management portal, leveraging AWS serverless architecture to achieve a 40% reduction in response time to performance outliers by automating real-time alerts • Led the development of ETL pipelines and data models to analyze clickstream data (1B+ rows), track customer service journey profiles and identify customer effort outliers to drive changes in customer returns and deliveries. Saved $15M in operational costs• Enhanced datawarehouse architecture and developed ETL pipelines with Glue orchestration to enable dynamic triggering of notifications that led to 20% improvement in user engagement• Standardized performance measurement for ~80k agents worldwide by applying the ‘balanced scorecard’ approach

Tucson, Arizona, Us

• Analyzed large datasets (3M+ rows), annotated with biological representations using Python to store in a graph database• Formulated NoSQL queries to identify relations and patterns in data by leveraging Neo4j for graph visualization• Confirmed hypotheses tests on genomic findings within a 5% significance level using statistical methods

Dallas, Tx, Us

• Pioneered development of an aircraft health monitoring platform to identify risk areas and adverse trends in real time• Conducted JAD sessions for gathering data requirements and GAP analysis to build Proof of Concept• Documented data flow diagrams, leveraged Python for language processing analytics• Demonstrated quantitative & qualitative analysis of operational data by creating SQL queries and Tableau dashboards

Cedar Rapids, Iowa, Us

• Managed AGILE development team adopting SCRUM methodology using JIRA; administered software build releases• Modeled software unit test cases using Python (5000+ lines of code), created code packages for avionics displays• Automated monthly data load (140GB) for simulators in Python to reduce manual effort by 80% and cycle time by 60%• Conducted project discussions & training with clients on multiple onsite trips to Cedar Rapids, Iowa• Developed SQL/PLSQL queries to pre-process heterogeneously sourced data (1M+ rows), reduced execution time by 70%• Derived new metrics into compliance and performance reports to suggest modifications to procedures and processes• Influenced reporting layer upgrade to Tableau, engineered a dynamic forecasting model to improve stability• Captured software specifications in the form of design documents and assisted in requirements translation

Master Of Science (Ms), Management Information Systems, General

B.E., Electronics And Communication

Science, Pcmb