Atmik Ajoy

Experience

1. Data Scientist 2

   Nike

   Jan 2024 — Present · Beaverton, OR, US

   Multimodal Feature Engineering for Profit-Driven Forecasting - Engineered product-context embeddings using purchase behavior, campaign metadata, and web-scraped trends in forecasting models,(Evaluated and using different LLMs) resulting in significantly improved accuracy and a projected $12M EBIT uplift.RAG-Driven Trend Detection for Forecast Enrichment - Designed a retrieval-augmented generation (RAG) system to inject real-time market signals and sentiment from scraped sources into Nike’s demand forecast pipeline—boosting adaptability to trend shifts.Built an LLM-powered system to classify consumer feedback into 205 business themes, driving insights into product innovation, marketing, and merchandising, and enabling faster, data-driven product and campaign adjustments by demographic, potentially decreasing product innovation time from 18 months to 10 monthsPySpark-Based Forecasting Pipeline Optimization - Re-engineered Nike’s legacy forecasting pipeline entirely in PySpark, optimizing ETL and parallel computation across 6B+ records to reduce compute time by 82%, enabling real-time forecasting and significantly accelerating strategic decision-making.Dynamic Time Series Model for Retail Optimization -Developed a hybrid time series model combining ARIMA, Exponential Smoothing, and regression techniques to optimize Nike’s store-level inventory forecasts—improving accuracy and reducing allocation errors by 7%, directly supporting retail execution.

Education

• PES University

  Bachelor's degree, Computer Science

  2018 — 2022

• University of Pennsylvania

  Masters of Science in Engineering in Data Science, Data Science