I use machine learning, SQL, and data pipelines to build advanced fraud detection systems and track complex risk patterns. I specialize in turning messy transaction data into real-time monitoring tools and actionable insights that protect financial infrastructure.

I started out driven by a curiosity for patterns and data signals. What began as an interest in how systems behave has grown into a dedicated focus on building data solutions that deliver clear, automated confidence in high-risk environments.
Today, my primary focus centers on fraud detection, risk analytics, and the modern data engineering pipelines that power them. I bridge the gap between raw data and real-world risk mitigation, ensuring clean architectures, high-precision detection models, and robust monitoring tools that support fraud operations teams.
Looking ahead, I am driven by one foundational question: Does this data help us catch bad actors and make better decisions? That is the standard of excellence I build toward—from data ingestion and SQL analytics to machine learning classification and real-time risk visibility.
My work focuses on the intersection of Fraud Detection, Risk Analytics, Predictive Systems, and Data Engineering.
Delivered 0.93 Precision and 0.82 Recall on a 0.17% imbalanced dataset, ensuring 93% of flagged transactions are confirmed fraud to mitigate financial loss.
Achieved a 51% MAE reduction by developing a real-time demand forecasting system on AWS to optimize dynamic pricing and fleet utilization.

Engineered a production-ready Medallion data platform using Apache Airflow and PostgreSQL to process 1M+ real-time transit records, maintaining a 76.1% data yield across 8 automated quality gates.
Tools and technologies I use to build production-ready data systems
Professional Impact & Results
Experience.experienceDesc1
Experience.experienceDesc2
Continuous growth in production ML and data engineering
Building neural networks for text classification, sentiment analysis, and language understanding with PyTorch and Transformers.
Implementing drift alerts, performance dashboards, and automated retraining triggers.
Exploring Feast and online feature pipelines for fraud detection and low-latency scoring.
Designing serverless workflows that orchestrate ML systems with reliability and traceability.
Building more resilient DAGs, improving observability, and applying data quality checks at scale.
Let's build something Amazing