Data Scientist: Causal, Fraud & Experimentation

Dileep Reddy Veesam

Frisco, TX +1-657-***-**** ******.*@*********.*** linkedin.com/in/veesamdileepreddy SUMMARY

Data Scientist with 4+ years of experience building predictive, causal, and anomaly detection models across payments, fraud risk, and cloud operations. Strong in Python, SQL, Spark, experimentation, causal inference, time series forecasting, XGBoost, and unsupervised learning, with experience analyzing global-scale datasets, productionizing ML models, and translating model outputs into measurable conversion, risk, and infrastructure decisions. EXPERIENCE

STRIPE Data Scientist Aug 2024 – Present

Payments, Experimentation & Machine Learning

• Engineered transaction and behavioral features from checkout events across 200 buyer countries using Python, SQL, and Spark, creating a shared feature base for conversion, revenue, and payment success analysis.

• Designed randomized holdout and A/B experiments with a 3 phase framework of power analysis, A/A tests, and pilots, validating randomization before launch so every treatment effect read was trustworthy.

• Modeled heterogeneous treatment effects with causal forests, pinpointing conversion uplift by geography, device, and industry, and verified CATE stability before results informed rollout decisions.

• Owned the experiment readout on incrementality, AOV, and revenue per session, isolating the segments that drove 7.4% higher conversion and 12% higher revenue and turning them into rollout recommendations.

• Shipped a payment method personalization model built on 100+ session and network signals, validated through online experiments that measured payment success and ranking lift by segment before rollout.

• Partnered with Product and Engineering to productionize model outputs, automating Python diagnostics that cut multi day segment analysis to hours while keeping statistical review and guardrails in place. UBER Data Scientist Jun 2023 – Jul 2024

Risk Analytics, Anomaly Detection & Machine Learning

• Aggregated rider, driver, payment, and device behavior into Hive feature pipelines across 50+ risk checkpoints, computed over 6 hour, 7 day, and 30 day windows to feed downstream fraud models.

• Trained unsupervised anomaly detection models on Michelangelo over 1,000+ entity level features, catching payment fraud, refund abuse, account takeover, and GPS spoofing to reduce fraud loss exposure.

• Calibrated anomaly and bucket thresholds with HAIFA feature importance, so every flagged entity carried an interpretable reason and investigators could triage cases faster.

• Monitored score drift, temporal shifts, and false positive patterns, pruning noisy variables to reduce false positives and sharpen prioritization of high risk entities.

• Deployed models via Python DSL and distributed training with Risk and Engineering, surfacing prioritized signals across 3 core workflows: rides, orders, and payouts for fraud loss and operational monitoring. Sage Softtech Associate Data Scientist Sep 2020 – Dec 2021

• Constructed Python and SQL pipelines ingesting logs, telemetry, and service traces from 10+ enterprise workloads, producing 50K+ validated time series observations for capacity analysis.

• Forecasted 30 day infrastructure demand with ARIMA, Prophet, and XGBoost, reaching 12% MAPE and benchmarking every model on RMSE and MAE before it was used in capacity planning.

• Applied K-means clustering and PCA to 20+ utilization features with silhouette validation, grouping workloads into steady state, seasonal, and burst driven patterns and improving compute sizing accuracy by 18%.

• Built regression models linking CPU, memory, request volume, and storage to infrastructure demand, exposing 15% over provisioned capacity and surfacing $600K+ in annual cost savings.

• Created Power BI and Tableau dashboards tracking 15+ operational KPIs and documented model assumptions, helping engineering and client teams make confident capacity planning decisions. SKILLS

Programming & Data: Python, SQL, Spark, Hive, Feature Engineering, Data Pipelines, Data Validation Machine Learning: XGBoost, Regression, Causal Forests, Anomaly Detection, K-Means, PCA, ARIMA, Prophet, Forecasting Experimentation: A/B and A/A Testing, Power Analysis, Causal Inference, CATE, Incrementality, Holdout Design Model Evaluation: RMSE, MAE, MAPE, Precision/Recall, Feature Importance, Explainability, Threshold Calibration Domain Analytics: Conversion and Revenue Analysis, Payment Personalization, Fraud and Risk Analytics, Capacity Planning Platforms & BI: Uber Michelangelo, Distributed Training, Model Productionization, Power BI, Tableau EDUCATION

California State University, Fullerton, CA Jan 2024 Master of Science, Computer Engineering

Contact this candidate