Deepika Jindal

About Deepika Jindal

With 9 years of experience leading large-scale machine learning initiatives, I specialize in delivering impactful solutions in recommendation systems, marketing analytics, NLP (BERT, GNNs), and time series forecasting, implementing Models- Neural networks, LSTM, Graph convolution networks, back propagations from scratch. Holding a Master’s in Data Science from Purdue University, I’ve earned 2 patents as a first inventor and presented research at the Amazon Machine Learning Conference.My expertise lies in leading cross-functional teams to execute high-impact projects across e-commerce, supply chain, and logistics. I excel at translating complex data insights into actionable business strategies and driving innovations that optimize performance and scalability.Key Skills:Deep Learning (Transformers, LSTMs, BERT)Machine Learning (XGBoost, Random Forest, GNNs)Time Series Forecasting, Fraud Detection, and NLPLeadership in Data-Driven Decision-MakingE-commerce & Supply Chain OptimizationTech Stack: Python, AWS, SQL, Hadoop, R, TableauExcited to bring my expertise in AI and machine learning to leadership roles, where I can continue pushing the boundaries of innovation and data-driven solutions.

Experience

1. Senior Applied Scientist

   Amazon

   Oct 2020 — Present · US

   Marketing Science- Developed agentic customer segmentation workflow for content personalization depth, with dynamic feedback customer features selection- Created multi agent workflow to develop content personalization instruction prompt and personalized content generation along with evaluation metrics- Led activation(first purchase) for B2B platform for Amazon $40Bn business, created science strategy for improving new customer purchase.Recommendations- Delevoped GNN based recommendations for business cross-sell, resulting in incremental OPS of $256MM WW. Lead end to end experimentation in realtime email settings and launched to production 2. Developed Search/ad keyword recommendations for cold-start problems shown incremental 110 bps in purchase rate. Created game theory based prescriptive ML model to draw most influential factors for purchase, with ~$250 MM YoY incremental revenue. Presented work in AMLC 20224. Launched activation propensity models to identify, measure and target customers with solid uplift, incremental $55M in reveProduct Compliance:1. Lead end to end Id validations project including for EPA (Environment Protection Agency), CEC, CDPR, FCC and 13 other use cases that validates the Id provided by the seller and the similarity of complain/non-complaint product with $8Bn impacted sales- Developed unsupervised textual similarity model that feeds on external database from compliance agencies for product compliance classification - Filed 2 patents (1 already Granted), on classification and resource planning as first inventor using unsupervised machine learning in the process- Presented featured talk in AMLC 2021 on product compliance classification using self-supervised 2. Lead at scale classification for all drugs- narcotics, Rx etc- drug categories, experimenting with BERT and variants,(S-BERT, DistillBERT, ROBERTA), Auto-gluon and conventional ML models, Sequential models RNN, CNN, LSTM

Education

• Purdue University

  Master of Science - MS, Data Science

• Deeplearning.ai

  Deep Learning Specialization, Deep Learning

  2019 — 2021

• Motilal Nehru National Institute Of Technology

  Bachelor of Technology, Electrical Engineering