
Anuska Roy
@anuskaroy
Data Scientist at Citi building a unified customer model for attrition, deposit, and investment predictions.
What I'm looking for
At Citi Bank, I built a Mixture-of-Experts model that predicts customer attrition risk, deposit propensity, and investment propensity in one architecture. I also used SHAP to explain model predictions and assessed feature stability across development and out-of-time datasets.
At IBM Research, I developed a multi-hierarchical RAG agentic system with LangGraph, OpenAI API, and Milvus. I also contributed to a diagnostic pipeline for silent failures in multi-agent systems, and co-authored research accepted at ACM ICPE 2026.
My research includes WFIL, a few-shot semantic segmentation framework using incremental learning and LoRA adapters, published at VISIGRAPP 2025. I also proposed a background-aware few-shot segmentation framework, with results reported on PASCAL-5 and COCO.
At Zeiss, I designed detector health evaluation tests for metrological X-ray CT systems. Earlier, as an Application Developer at Accenture, I built and maintained .NET Core microservices and deployed them through Azure DevOps CI/CD pipelines.
Experience
Work history, roles, and key accomplishments
Designed and deployed scalable AI architectures for production-ready solutions. Built a Mixture-of-Experts model for multi-task customer classification and delivered model explainability using SHAP.
Designed and implemented detector health evaluation tests for metrological X-ray CT scan systems in compliance with ASTM standards. Analyzed detector performance and reliability metrics to identify failure patterns.
AI Research Intern
IBM Research
May 2025 - Aug 2025 (3 months)
Developed a multi-hierarchical RAG agentic system using LangGraph and OpenAI API, and built a diagnostic pipeline for silent failure detection in multi-agent systems. Co-authored research accepted at ACM ICPE 2026.
Built and maintained microservices using .NET Core with full test coverage and deployed via CI/CD pipelines on Azure DevOps. Gained foundational experience with large-scale software systems and cloud infrastructure.
Education
Degrees, certifications, and relevant coursework
International Institute of Information Technology, Bangalore
Masters by Research, Data Science
2023 -
Grade: 3.5/4
Pursuing Masters by Research in Data Science with a thesis on scaling generalized few-shot semantic segmentation to large-scale novel class addition.
Kalinga Institute of Industrial Technology (KIIT)
Bachelor of Technology, Electrical Engineering
2017 - 2021
Grade: 9.0/10
Completed Bachelor of Technology in Electrical Engineering with a grade of 9.0/10.
Availability
Location
Authorized to work in
Salary expectations
Social media
Job categories
Interested in hiring Anuska?
You can contact Anuska and 90k+ other talented remote workers on Himalayas.
Message AnuskaGet matched with your dream remote job
Sign up now and join over 250,000+ remote workers who receive personalized job alerts, curated job matches, and more for free!
