JASPREET SINGH

Advance Data Scientist, Matrixcare

ai.jaspreetsingh@gmail.com • +91-9780-9535-56 •

LinkedIn • Github • Blog

Executive Summary

Senior Data Scientist with 8+ years of experience in building advanced analytical tools for Fortune100 companies like Walmart, JP Morgan, Dubai Department of Economy and Tourism, Resmed-Matrixcare, etc. Responsible for project planning, execution and team management

Work Experience

Advance Data Scientist | Matrixcare-ResMed

September 2025 - Present
Fallrisk

  • Skilled-nursing facilities needed earlier identification of residents at risk of falling, using clinical data spanning diagnoses, medications, mobility, vitals and fall history.
  • Combined clinical rules with Random Forest predictions and SHAP explanations to generate resident-level risk scores; modernised the ML pipeline using XGBoost and updated MDS Section GG features.
  • Reduced the model artifact from 565 MB to 125 KB—approximately 4,500× smaller—with the upgraded model deployed in shadow environments pending clinical hybrid-score recalibration.
  • Tools: Python, FastAPI, scikit-learn, XGBoost, SHAP, AWS S3

Depression Risk

  • Depression indicators in long-term care could remain unnoticed between scheduled assessments, creating a need for continuous monitoring across clinical records and daily-living activities.
  • Integrated a DSAIL-hosted CatBoost model with EHR workflows, using SHAP-selected diagnoses and ADL patterns alongside clinical rules, OAuth2 authentication and PostgreSQL response caching.
  • Enabled explainable resident identification through a seven-factor framework, flagging residents with four or more indicators to support targeted depression screening and care-plan review.
  • Tools: Python, FastAPI, CatBoost, SHAP

Senior Data Scientist | Walmart

Dec 2024 - September 2025
Deduplication

  • Duplicate seller-listed SKUs on the e-commerce platform created redundant product listings, requiring identification and resolution to improve the shopping experience.
  • Developed a product deduplication solution using Sentence Transformers / SBERT for textual embeddings and CLIP for image embeddings, applying cosine similarity to identify matching titles, descriptions and product images.
  • Applied deduplication logic to reduce listing redundancy and support a better customer experience and higher purchase likelihood.
  • Tools: Python, Google Vertex

Product Attribute Enrichment

  • Product catalogues required more complete and consistent attribute information to improve item coverage and conformity.
  • Developed a GPT-4o-based generative AI solution to automate product attribute enrichment at scale.
  • Achieved item coverage and conformity scores exceeding 95% across 13 million products.
  • Tools: Python, OpenAI, Google Vertex

Senior Data Scientist | Maaloomatiia - PCS

Sep2023 - Dec 2024
Consumable Energy Prediction

  • Dubai Metro faced challenges in accurately forecasting water, AC & DC electricity, and cooling consumption for the Green and Red lines, leading to frequent resource shortages and excesses
  • Leveraged N-BEATS algorithm to enable Dubai Metro with insights on precise & reliable consumption demands
  • Yielded savings of approximately $80K annually by reducing stagnation and resource outages, while simultaneously meeting sustainability goals
  • Tools: Python, Dataiku, NeuralForecast

Cab Complaint Classification

  • The Road and Transport Authority (RTA) customer service webpage allowed users to share complaints about cab services and RTO experiences, which were manually routed to the appropriate department, service, or touchpoint by the back-end team for resolution
  • Delivered a Llama3-8b-supported NLP model combined with an XGBoost classifier that can effectively handle multi-hierarchical classification of user complaints for quicker resolution
  • The model resulted in a 66.67% reduction in complaint acknowledgment lead time while maintaining 72%-89% accuracy in model performanceF
  • Tools: Python, Dataiku, Ollama, Llama3-8b, XGBoost

Overnight Visitor Arrivals and Spend Prediction

  • Dubai government's Department of Economy & Tourism (DET) wanted to understand the factors that lead to changes in overnight visitor footfall and how to predict accurate arrivals from each source market
  • Developed an XGBoost-based solution equipped with What-If analysis tool to understand the factors leading to overnight visitor arrivals and the economic impact they generate for the Dubai government
  • The model was used by the CEO of DET and government body members to re-evaluate global marketing expenditure and experiment with other tourism-related factors to channel demand based on needs
  • Tools: Python, Dataiku, Databricks, SciKit-learn, NBEATS, PatchTST, Streamlit

Social Media Analysis and Insights

  • Dubai's immigration services faced challenges in gauging public opinion, identifying key themes, and responding to emerging issues
  • Implemented advanced NLP techniques for sentiment analysis, LLM-based text categorization, and summarization to extract insights from social media posts and public feedback, enabling efficient data-driven decision-making
  • Improved operational efficiency by tracking public service KPIs and providing timely insights into public sentiment, helping the GDRFA address issues more effectively, prioritize service improvements, and enhance customer satisfaction
  • Tools: Tech Stack Used: Python, Dataiku, Ollama, Llama2-3b, Generative AI

SKU Purchase Prediction

  • One of the Fortune 500 sportswear retail organizations faced challenges in accurately predicting SKU purchases, leading to frequent stockouts and overstock situations, causing significant revenue losses and storage costs
  • Developed a Temporal Fusion Transformer (TFT)-based advanced time series model to deliver precise and reliable SKU purchase forecasts
  • The predictive model implementation resulted in 40% reduction in stockouts and overstock situations, saving approximately $560k annually and enhancing inventory management and operational efficiency
  • Tools: Tech Stack Used: Python, Datarobot, TensorFlow, Keras

ATM Cash Withdrawal Prediction

  • Arab Bank wanted to improve the customer service satisfaction by reducing the Out-of-Cash scenarios in their ATM through out the Egypt and Jordan network
  • Developed a cash withdrawal demand prediction model using Random Forest regressor which assisted operations teams to proactively look out for potential Out-of-Cash situations
  • The ATM Cash demand forecasting model was used by both operation team and ATM vendors to monitor and replenish the flagged ATMs
  • Tools: Tech Stack Used: Python, XgBoost, LGBM

Senior Data Scientist | J.P. Morgan

Feb2022 - Sep2023
Attention & Deep Learning Based Email Classifier

  • The Service & Implementation team had to go through hundreds of incoming emails a day to comprehend and tag the relevant team for issue resolution
  • Implemented an attention-based neural network using TensorFlow to understand and classify emails into appropriate categories and teams
  • The solution reduced the manual labor required to identify and tag emails to teams for more than 150 mailboxes
  • Tools: Tech Stack Used: Python, Tensorflow, JupyterLab, AWS

Classification Based Analogous Client Finder

  • Currently numerous billing platform are used by JP Morgan's commercial banking unit, tailoring to client's specificity which adds extra burden on operations and maintenance cost
  • Proposed XGBoost based classification solution lead to operational savings in production deferral costs through strategic client migration
  • Provided solution was used to segregate for migration plan
  • Tools: Tech Stack Used: Python, Tableau, Scikit-Learn

Latent Dirichlet Allocation Based Request Categorization

  • Eliminated Onboarding & Servicing team's invisibility to unseen trends and change in key issue subjects
  • Developed Latent Dirichlet Allocation based topic modeling & pattern recognition, leading to automated category and subject identifier
  • Transformed solution was adopted by Commercial Banking's client service team to understand change in client issues to improve process optimization
  • Tools: Tech Stack Used: Python, NLTK, Sklearn

Data Scientist | Mu Sigma

Dec 2018 - Jan 2022
Random Forest Based Manufacturing Halts Reduction

  • Emirates Global Aluminum (EGA) was facing multiple frequent production halts due to machinery faults which would last for more than weeks and cost multiple millions dollar in loss of opportunity
  • Assisted EGA to reduce unplanned maintenance shutdown, production halts and improve equipment life cycle by predicting potential machinery failures well in advance
  • Proposed Random Forest based predictive solution lead to operational savings of over millions annually in production deferral costs
  • Tools: Tech Stack Used: Python, Scikit-Learn, Azure DataBricks, Tableau

Computer Vision & Deep Learning Based Brick & Mortar Customer Analysis

  • Understanding offline customer's behavior patterns to make better decisions in store operations (staff management, product placements, etc)
  • Implemented in-store video analytics solution using Single shot multibox detector (SSD) based YOLO V3 solution
  • Delivered hour level data about customer entry, exit & in-store count, number of aisle visits, traditional checkout counter count, Scan-&-Go counters encouragement and others
  • Tools: Tech Stack Used: Python, YOLO v3, OpenCv

Natural Language Processing Based Early Trends Detector

  • Eliminated sourcing & procurement team's invisibility to unseen trends
  • Developed Natural Language Processing based model lead to 3 fold decrease in Out-of-Stock scenarios
  • Transformed solution was adopted by clients as their official banner product for 2019 Black Friday Sale
  • Tools: Tech Stack Used: Python, NLTK, Tableau

HOBBY PROJECTS

Technical Skills

Machine Learning: Linear Regression, Logistic Regression, SVM, KNN, Decision Tree & Random Forest, Ensemble Models
Deep Learning: Convolutional Neural Network(CNN), RNN, Attention, BERT, Natural Language Processing, Computer Vision
Tools: TensorFlow, Keras, Pandas, Numpy, Matplotlib, Seaborn, scikit-learn, ARIMA, Prophet, NLTK, OpenCV, Yolo V3, Databricks, Dataiku, Datarobot, Jira, Kanban Board
Misc: Git, GitHub, Python, SQL, Probability, Statistics

Education

Achievements

Rising Star- Commercial Bank - JP Morgan 2022
MAGniverse: In-house Decision Making Framework deliverd from ideation to production