Capital One

> Leading the first party fraud detection R&D. This involves user behavior sequence modeling, working with financial traits, graph-based features and customer/merchant embeddings to predict propensity to commit fraud.

> Working with multiple stakeholders (business, engineering, model risk) to productionize ML models and expanding their use across multiple Upmarket card products.

> Delivered $50M per year in savings against fraud losses.

Lead Machine Learning Scientist

Build models for first party fraud detection in credit cards.

> Research, train, productionize, and monitor ML models for various customer segments. Saving $XX millions in fraud losses.
> Few tasks for model improvement - Hyperparameters tuning, resolve class imbalance using weights, implement custom loss function that closely resembles business metric, information theoretic feature selection.
> Present lessons learned at internal and external events.
> Work with H2O.AI to implement and test new functionalities in their open-source library.
> Skills: Python, Spark, H2O, Scikit-Learn, XGBoost, AWS.

Senior Machine Learning Scientist

Cerner Corporation

Within Population Health, I helped Cerner support organizations seeking to improve the quality of care, operate more efficiently, improve financial management and engage people in managing their healthcare.

> Worked on ingestion platform to extract, transform, and load electronic medical records (from disparate sources) into HBase using MapReduce.
> Lead developer to implement in-order batch processing of entity types, prerequisite to onboard UK clients under a strict timeline. Use case that was previously unsupported by existing framework.
> Sole Reference-Record engineer in an across teams project to migrate Population Health clients to CDH5 cluster for faster data processing and less resource contention.
> Skills: Java, Hadoop, HBase, Apache Crunch.

Senior Software Engineer

General software engineering.

> Developed crash analyzer for FirstNet to compare WinDbg crash dumps against cataloged data.
> Resolved 760 service requests, 45 post-upgrade crashes across 193 clients. 50% more than expected target.
> Received "Cerner All-Star" award for utmost client satisfaction. This award is given to top 2% associates who go above and beyond normal role expectations.
> Skills: C++, SQL.

Software Engineer

Kansas State University

> NSF funded research aimed to construct statistical models to characterize Gene Regulatory Networks involved in plant responses to abiotic stresses.
> Explored and published a "motif-based machine learning approach" to identify regulatory elements in plant (Arabidopsis) genome.
> Performed feature selection (based on information gain & ranker's search) and feature abstraction (based on hierarchical agglomerative clustering) to identity 182 (out of 578) relevant features (or motifs).
> Grouped motifs into 72 biologically significant concept-clusters to increase classifiers' AUROC by 11%.
> Predicted (with 88% accuracy) transcription pattern for genes in Arabidopsis using transcription factor binding sites from plant databases as features.
> Executed numerous training experiments with best possible features and their vector representations in order to generate robust ML models.
> Skills: Java, Weka.

Graduate Research Assistant, Machine Learning & Data Science Lab

At Capital One, I use machine learning and data science to detect and prevent fraud and abuse in credit cards.

Industry experiences:
> 7+ years of r&d in Financial Services. 
> 7+ years of software & data engineering in Healthcare.

You may reach me at vishalbahirwani@gmail.com.

Taplio

Vishal Bahirwani's Linkedin Analytics

Want to drive more opportunities from LinkedIn?

Taplio

.css-1jgreyo{background-image:linear-gradient(284deg, rgba(39,43,72,1) 0%, rgba(33,47,183,1) 100%);color:transparent;-webkit-background-clip:text;background-clip:text;}Vishal Bahirwani's Linkedin Analytics

Want to drive more opportunities from LinkedIn?

Vishal Bahirwani's Linkedin Analytics