Varshith Tarani

About Varshith Tarani

As a data-driven and detail-oriented Generative AI Engineer, I specialize in designing and deploying AI solutions that transform complex data into intelligent, scalable systems. With a Master’s in Business Analytics & Information Systems from Concordia University Wisconsin, I combine strong analytical foundations with hands-on experience in large language models, predictive modeling, and AI-driven decision support systems.My professional journey began in my family’s jewelry business, where I worked closely with operational and financial data to improve forecasting, validate insights, and support growth through data-backed strategies. This early exposure to real-world business problems shaped my ability to build AI solutions that are practical, measurable, and aligned with business outcomes.As I moved into analytics and AI-focused roles, I gained experience building and optimizing LLM-powered applications, data pipelines, and intelligent workflows. I have worked extensively with tools such as Python, SQL, Power BI, Tableau, and machine learning frameworks to develop predictive models, automate data processes, and enable smarter decision-making at scale.My experience as a Business Development Analyst at VS Jewels and a Content Moderator at Accenture strengthened my skills in data governance, automation, model evaluation, and cross-functional collaboration—all critical for building reliable and responsible GenAI systems in production environments.Core Skills: Generative AI & LLMs | NLP | Predictive Modeling | Python | SQL | AI Pipelines | Data Visualization (Power BI, Tableau)| Business IntelligenceGoal: To build and scale Generative AI solutions that deliver real business impact, improve decision-making, and drive innovation through responsible and data-driven AI systems.

Experience

1. Ai Engineer

   Stripe

   Jun 2025 — Present · US

   I build production ML systems that detect risky payment behavior at scale and reduce financial exposure.Train and optimize ensemble models (XGBoost, Random Forest) on 500K+ transactional and behavioral records, improving accuracy from 82% to 91%.Engineer time-based and behavioral features to surface early fraud signals and usage anomalies.Contribute to a 28% reduction in fraud losses (~$1.2M annually).Own the full ML lifecycle — data processing (Pandas), model development (Scikit-learn, TensorFlow), validation, deployment, and monitoring.Deploy real-time inference services (<200ms latency) on AWS (SageMaker, EC2, S3, Lambda) with CI/CD automation and model versioning.Maintain 99.5% production reliability through monitoring and drift detection.Partner with risk and compliance teams to ensure model outputs are interpretable and production-safe.I focus on building models that are not just accurate, but resilient, scalable, and financially impactful.

Education

• Concordia University-Wisconsin

  Master's degree, Information Technology

• Osmania University, Hyderabad

  Bachelor's degree, Business/Corporate Communications

  2017 — 2020