Vihitha R

About Vihitha R

AI/ML Engineer with 3+ years of experience in designing, developing, and deploying machine learning models, large language models (LLMs) like LLaMA and Mistral, and AI-driven solutions. Skilled in supervised/unsupervised learning, NLP, deep learning, and the end-to-end ML lifecycle, including data preprocessing, feature engineering, model evaluation, and deployment. Experienced in MLOps, model optimization, and cloud deployment on AWS, Azure, and GCP, leveraging Python, TensorFlow, PyTorch, SQL Server, and PostgreSQL. Hands-on expertise in LLM frameworks such as LlamaIndex, RAG, fine-tuning, and prompt engineering to deliver context-aware AI solutions.

Experience

1. Ai Ml Engineer

   JPMorganChase

   Jan 2025 — Present · US

   Built advanced fraud detection pipelines using TensorFlow, BERT, and GPT-4, significantly enhancing transaction monitoring by reducing false positives by 27% and improving overall anomaly detection accuracy, ensuring safer financial operations. • Implemented end-to-end MLOps workflows leveraging MLflow and Vertex AI, automating model training, validation, and deployment processes, which reduced release cycles and increased team productivity. • Developed sophisticated Named Entity Recognition (NER) models to extract key entities such as customer IDs, merchant names, and transaction categories, accelerating fraud case resolution by 30% and improving operational efficiency in investigations. • Intended Retrieval-Augmented Generation (RAG) pipelines using LangChain and LlamaIndex, allowing financial advisors to efficiently query vast volumes of policy and compliance documents, which cut manual lookup time by 40% and streamlined advisory workflows. • Experimented with Generative Adversarial Networks (GANs) to generate synthetic transaction data, strengthening fraud detection models for rare-event scenarios without compromising sensitive customer information, enhancing model robustness. • Established AI-powered dashboards in Tableau, delivering real-time insights into fraud alerts, model performance metrics, and customer engagement trends, enabling executives to make informed, data-driven decisions. • Generated large language models (LLMs) such as LLaMA and integrated them with LlamaIndex to build intelligent conversational banking assistants, increasing customer self-service adoption by 22% and reducing operational workload on support teams.

Education

• Sri Chaitanya College of Education

  Intermediate

• DR. K.K.R. GOWTHAM COLLEGE OF EDUCATION

  SSC

• Vignan's Foundation for Science, Technology & Research

  Bachelor's degree, Computer Science

• UMass Boston

  Master's degree, Computer Science

  2025