Sudeeksha Vandrangi

About Sudeeksha Vandrangi

Hi there! I’m Sudeeksha, a data scientist with a proven track record of leveraging machine learning and analytics to drive measurable impact.At Tata Steel, I reduced defective slab production by 15% through root cause analysis and implemented a logistic regression model with 90.43% accuracy. At Lennox International, I optimized HVAC scheduling by analyzing over 9M+ data points using Apache Spark, PyTorch-Forecasting, and Databricks. I also implemented real-time feedback systems via Azure Communication Services, streamlining decision-making processes.With expertise in Python, SQL, Apache Spark, PyTorch, and Azure ML Studio, I excel at developing scalable solutions for complex problems and collaborating with cross-functional teams. I’m actively seeking full-time opportunities as a data scientist to bring data-driven insights to high-impact projects. Let’s connect to explore how I can contribute to your team’s success!

Experience

1. Data Scientist

   Peerlogic

   Sep 2025 — Present · Scottsdale, AZ, US

   Leading the full lifecycle of the Propensity To Show (PTS) feature, including Product Requirements Document (PRD) writing, data design, feature engineering, evaluation, deployment, and ongoing monitoring. Building production feature pipelines using Python, BigQuery SQL, Vertex AI, and Django REST, ensuring leakage-safe feature computation and consistent inputs across API, Insights, and Pipey services. Improving model training speed by running GPU-accelerated jobs on Vertex AI, reducing training time from about 11 hours to under 40 minutes and accelerating iteration during development. Designing the inference architecture connecting Postgres, BigQuery and a containerized Django REST scoring API to support appointment-level predictions across product workflows. Implementing appointment-triggered workflows using RabbitMQ so Pipey can hydrate appointment and patient data, compute features, call the scoring endpoint, and store results for downstream logic. Collaborating with Product, Sales and Engineering cross-functionally, to align data flows, model requirements, and integration points, reducing cross-service friction and making rollouts more reliable. Adding tests, documentation, and architecture diagrams that clarify system behavior and improve engineering handoffs for future contributors.

Education

• National Institute of Technology Rourkela

  Bachelor of Technology, Materials Engineering

• National Institute of Technology, Rourkela

  Bachelor of Technology - BTech, Metallurgical and Materials Engineering

• Stanford University

  Code in Place- Student Volunteer

  2020 — 2020

• Arizona State University

  Master of Science, Data Science

• Arizona State University

  Master of Science - MS, Data Science Analytics and Engineering