Senior Data Scientist in Personalization & Causality

Ayesha Khan

Senior Data Scientist Personalization, Ranking & Causal Inference

Monroe Township, NJ, 08831. 325-***-**** ******.*********@*****.*** PROFESSIONAL SUMMARY

Senior Data Scientist with nearly 8 years of experience building production machine learning systems for personalization, ranking, and causal measurement. Developed recommendation and ranking models on large-scale transaction and behavioral data that lifted click-through, conversion, and 90-day retention. Designed 15+ A/B and incrementality experiments and applied causal inference (matching, difference-in- differences, uplift modeling) to quantify campaign ROI for brands and partners. Works with engineers to ship models into production, from feature stores to batch and real-time scoring and monitoring, and owns projects end to end from problem framing to post-launch evaluation. Known for turning complex model results into clear recommendations for product, sales, and executive audiences. CORE SKILLS

•Machine Learning: Recommendation systems, learning-to-rank, propensity and uplift models, collaborative filtering, embeddings, XGBoost, PyTorch, scikit-learn.

•Experimentation & Causal Inference: A/B testing, incrementality testing, matching, difference-in- differences, uplift modeling, power analysis.

•Languages & Data: Python (pandas, NumPy), SQL, Spark, Airflow, dbt, feature stores.

•Cloud & MLOps: AWS (S3, SageMaker, EMR), batch and real-time scoring, model monitoring, REST APIs, Docker, Git, CI/CD.

•Domains: Fintech, e-commerce, transaction-level and event-driven data.

•Tools: Tableau, Jupyter, MLflow.

PROFESSIONAL EXPERIENCE

Senior Data Scientist

JK Techhub

12/2024 – Present

•Led development of a personalization and ranking model on transaction and behavioral data, improving click-through rate by 12% and conversion by 8% across 3M+ users.

•Designed and analyzed 15+ A/B and incrementality experiments, and standardized sample-size, guardrail- metric, and readout practices adopted across the team.

•Applied causal methods (matching, difference-in-differences, uplift modeling) to quantify campaign ROI, informing $2M+ in annual budget decisions.

•Built an attribution and measurement system that estimates the incremental value of campaigns, giving brands and partners clear ROI reporting and cutting reporting turnaround from 2 weeks to 3 days.

•Improved cold-start recommendations for new users and merchants with content-based features and embeddings, lifting first-session engagement by 9%.

•Designed a feature store with versioned, reusable features for personalization models, reducing duplicate feature work across 4 teams.

•Implemented real-time scoring for ranking models with under 100 ms latency, working with engineering on API deployment and monitoring.

•Set up automated data-quality checks and model-performance dashboards, catching pipeline issues 70% faster.

•Optimized ranking for retention as well as short-term clicks, improving 90-day retention by 5%.

•Presented experiment results and model tradeoffs to executives and external partners, earning approval for 3 roadmap investments.

•Shipped 6 models into production, from problem framing and data exploration through deployment and post-launch evaluation.

Data Scientist

Monte Carlo

07/2021 – 11/2024

•Developed propensity and churn models in Python (XGBoost, scikit-learn) that improved retention by 6% and targeting precision by 15%.

•Built collaborative filtering and embedding-based recommendation prototypes that improved offline NDCG by 14% over the baseline.

•Created customer segmentation and merchant-affinity features from transaction data, lifting campaign response rate by 11%.

•Ran 10+ A/B tests end to end, from hypothesis and design through analysis and rollout recommendation.

•Introduced model evaluation standards (holdout design, calibration checks, drift monitoring) that reduced post-launch performance issues by 30%.

•Worked with data engineers to move models from notebooks to scheduled Airflow batch jobs, cutting manual runs by 80%.

•Translated model results into dashboards and summaries for non-technical stakeholders, speeding up decision-making on campaigns.

Data Analyst

Selby Jennings

12/2018 – 06/2021

•Analyzed customer and transaction data with SQL and Python to uncover trends that drove 10% revenue growth in a key segment.

•Built cohort and funnel analyses to identify drop-off points, improving conversion by 7%.

•Defined and standardized core business KPIs (conversion, retention, lifetime value) used across 5 teams.

•Cleaned and validated datasets from multiple sources, reducing reporting errors by 25%.

•Built Tableau dashboards used by 25+ stakeholders for weekly decision-making.

•Supported early A/B tests with sample-size calculations and result summaries, building the basis for the experimentation program.

SELECTED PROJECTS

Offer & Product Recommendation Engine

Two-stage system (candidate generation with embeddings, then gradient-boosted ranking) on user– merchant interaction data. Evaluated offline with NDCG and MAP, then online through an A/B test. Stack: Python, XGBoost, PyTorch, Spark, AWS.

Incrementality & Campaign ROI Measurement Framework Measurement framework using matched control groups and difference-in-differences to estimate true incremental spend, replacing last-touch attribution. Delivered partner-facing ROI reports that improved partner renewal by 10%.

Uplift Modeling for Targeting

Uplift models that identify the users most likely to respond to incentives, reducing incentive cost by 18% while maintaining conversion. Validated with a holdout experiment. Production ML Pipeline & Monitoring

Feature-store-backed training and scoring pipelines (Airflow, dbt) with automated data-quality checks and drift alerts, cutting model failures by 60%.

EDUCATION

Temple University. Philadelphia, PA.

Bachelorʼs Degree.

Contact this candidate