Yaswanth J

About Yaswanth J

With a strong foundation in AI and machine learning, I specialize in building and deploying advanced Generative AI solutions to drive innovation in the banking sector.• Designed and fine-tuned Large Language Models (LLMs) like T5, GPT, and BERT for personalized customer communications, fraud detection, and risk assessment.• Built and deployed scalable AI workflows using Vertex AI, AWS SageMaker, and Azure OpenAI, enabling seamless automation of critical business processes.• Developed MLOps pipelines for efficient model deployment, versioning, and monitoring, ensuring compliance with regulatory requirements and industry standards.• Leveraged frameworks like LangChain and Retrieval-Augmented Generation (RAG) to enhance customer query resolution and document summarization.• Integrated APIs and cloud platforms, including AWS, Azure, and Snowflake, to deliver secure, scalable, and efficient AI-driven banking applications.I hold a Master of Science degree in Computer Science from Fitchburg State University, where I developed expertise in data science, machine learning, and artificial intelligence. Throughout my academic journey, I refined my skills to apply advanced concepts and techniques in these domains. My primary objective is to leverage my education and skills to help organizations make informed, data-driven decisions that align with their strategic goals. Additionally, I am a certified Data Engineer, validating my proficiency in cloud-based technologies. With a solid educational foundation and practical certification, I am well-equipped to contribute effectively to the dynamic field of data engineering and analytics.

Experience

1. Agentic Ai Developer

   Verizon

   Dec 2025 — Present · NJ, US

   Currently focused on agent performance optimization and infrastructure development for an enterprise-grade network analytics platform. Leading initiatives in multi-tier cache architecture design and implementation, where I’ve built a 3-phase caching system (tool-level, SQL-level, and BigQuery result-level) that achieves latency reduction for repeat queries and 92% improvement for similar queries. Developed a comprehensive RESTful API with 7 endpoints for cache management supporting granular invalidation strategies (tool-based, pattern-based, age-based) complete with authentication, rate limiting, and real-time metrics monitoring. Conducted in-depth LLM performance analysis and bottleneck identification, documenting 4-7x infrastructure degradation through systematic trace analysis. Currently building privacy-compliant query optimization and validation systems using pattern-based entity extraction and static templates to improve user query quality without storing sensitive data. Working extensively with Python, FastAPI, Google BigQuery, Google ADK (Agent Development Kit), LangChain frameworks, RESTful API design, and Google Cloud Platform (GCP). Focus areas include AI agent performance optimization, distributed caching strategies, LLM prompt engineering, system observability, and cost reduction through intelligent infrastructure

Education

• University of Bridgeport

  Master of Science - MS, Computer Science

• G. Pulla Reddy Engineering College

  Bachelor's degree, Mechanical Engineering

  2013 — 2017

• Fitchburg State University

  Master of Science - MS, Computer Science