Kevin G. Email & Phone Number

United States

Led a team of interns to construct a database and develop a RAG chatbot using Llama3-70b and Langchain framework with reinforcement learning and feedback loop to improve customer question answering, resulting in a 20% reduction in customer service costs.Developed and optimized OCR algorithms for handwritten menu recognition using Tesseract and OpenCV, increasing text recognition accuracy by 25%.Created customer profiles using clustering on customer behavior datasets—improving content recommendations and increasing revenue by 20%.Built a Python-based recommendation engine using a DeepFM model on customer segmentation and restaurant online order data—boosting customer dish order conversion rates by 4%.Developed and optimized XGBoost predictive models with hyperparameter tuning to forecast user engagement—raising retention by 15% through personalized promotions.Developed interactive Tableau dashboards to communicate user-defined key metrics in restaurant operations to stakeholders, enhancing decision-making efficiency by 25%.Configured an event-driven pipeline using AWS S3, SNS, SQS, and Lambda for message queuing and orchestration, automating ingestion and ensuring real-time data availability. Implemented monitoring and logging to track performance and maintain uptime, reducing processing time by 40% and increasing throughput by 30%.Implemented and orchestrated containerized data pipelines using Docker and Kubernetes, deploying Redshift and Mage for scalable data processing. Utilized Docker to streamline pipeline management and deployment, improving data processing efficiency by 40% and reducing ETL time by 30%.

Greater Boston

Drove a 3% increase in customer retention rates for clients in the US by employing an autocorrelation distributed lag model on 20-year longitudinal data to elucidate the impact of inflation on total expenditure, guiding strategic pricing decisionsEnhanced client's strategic selection of bands for collaboration and optimal concert scheduling by identifying the economic impact of bands' tours on local economies through synthetic control analysis, utilizing 20 TB of data from 23,000 bands/singers and leveraging parallel computing techniques as a student consultantUsing MySQL to compute the customer loyalty in a 20 year panel of retail stores

Greater Boston

Collaborated with a team of 5 data scientists to diagnose impact of optimizing marketing channels by a mixed effect model in Python using 40TB transaction data, generated $270M increase in digital revenue by refining budget allocation.Reduced inventory cost by 30% in a hyperparameter-tuned LSTM model, improving forecast accuracy by 25% over the original ARIMA model to enhance demand forecasting and supply chain management for digital sales.Curtailed user churn rate by 20% by constructing a predictive model using LightGBM based on customer profile and RFM data to target at-risk customers and increase their shopping frequency.Developed and operationalized AWS data pipelines (API Gateway, Lambda, S3) for diverse tables, ensuring consistent data processing and timely delivery.Streamlined data ingestion from Google Pubsub using AWS Kinesis Firehose and API Gateway, enabling real-time analytics and boosting data accessibility by 30%.Introduced the use of DBT for ETL (Extract, Transform, Load) development, leading to 20% decrease in testing requirements and 50% enhancement in code integrity.

Boston, Massachusetts, United States

Built ETL pipeline using MS SQL Server and AWS RDS, reducing data retrieval time by 40% for analyticsAutomated data migration to AWS S3 with AWS DMS, increasing data availability by 30%Created Delta Live Table pipelines in Databricks, enabling real-time data updates for personalized learning content

Shanghai, China

Coordinated the transition of historical retail market data from DynamoDB to Snowflake, guaranteeing smooth data migration and reducing downtime to a minimum.Deployed cross-account data ingestion pipelines, facilitating smooth data transfer between S3 buckets located in distinct AWS accounts.Reduced user churn over 10% by analyzing operations in client relationship management and new product testing through competitive analysis.Decreased the operational cost to 60% by investigating operation status of three creative new food and beverage brands; results were included in report about the evolution of new retail businesses.

Xiamen, Fujian, China

Wrote a report on Medical Policy Reform that provided suggestions to the government on impacts to social welfare. Led a team of 12 to complete a causal inference analysis on the medical big data in Xiamen from 2000 to 2021.

Shanghai, China

Conducted benchmark analysis against leading firms to optimize operational efficiency by over 10%Automated data collection pipeline using selenium and Python, increasing data retrieval speed by 70%Built revenue forecasting model using XGBoost, reducing SMAPE to 3% through hyper-parameter optimization

Beijing, China

Developed a quantile regression neural network model on 6,000 days of trading data, achieving a 20% annualized returnBuilt a CNN model for alpha generation, resulting in a 15% improvement in prediction accuracyCreated a sentiment analysis model for intra-day trading, leading to a 15% increase in trading signal accuracy

Master Of Science - Ms, Analytics, 4.0

Activities and Societies: Hassenfeld Immersion Program Fellow Rachel McCulloch Fellowship

Master Of Arts - Ma, International Economics And Finance, 4.0

Activities and Societies: GMIC 13th Stock Pitch Competition Second Place Brandeis Consulting Association Case Competition Finalist

Bachelor’S Degree, Mathematical Statistics, 3.8/4.0

Activities and Societies: Chairman of a volunteer committee Dean’s list Courses taken: Introduction to Probability, Political Economics.

Bachelor Of Science - Bs, Physics, 3.8/4.0

Activities and Societies: Scholarship for Outstanding Students (2019-2022)(top 5%) Scholarship for Social Activities(2020) Courses taken.