Sahaza Shakya

Senior Software Engineer / AI @ Verizon

About

I build backend systems that scale and AI systems that actually work in production.Over 6 years, I've gone from owning high-concurrency Java modules at Southwest Airlines to architecting multi-agent AI frameworks at Verizon that cut operational overhead by 40%. The through line: I don't just implement.I own outcomes.What I do today at Verizon:▸ Design Multi-Agent Collaboration (MAC) systems using Spring AI + ▸ LangGraph that automate complex, multi-step business logic at enterprise scale▸ Build RAG pipelines with Hybrid Search (pgvector + semantic) that pushed internal retrieval precision from ~67% to 94%▸ Ship AI Guardrail layers that intercept LLM outputs in <30ms — protecting customer-facing interactions from hallucinations and policy violations▸ Create Human-in-the-Loop feedback loops that compress prompt iteration cycles and reduce regressions by 60%My stack: Java 17/21 · Spring Boot 3.x · Spring AI · LangGraph · LangChain · Python · AWS Bedrock · Pinecone · pgvector · Docker · Kubernetes · Terraform · Kafka · GrafanaI'm most valuable to teams that are serious about moving Java-based backends into AI-enabled products not as a side project, but as a core capability.Open to Senior Software Engineer opportunities. Let's talk.

Country

United States

City

Dallas-Fort Worth Metroplex

Industry

Information Technology & Services

Skill

Generative AI, Java, Spring Boot, Retrieval-Augmented Generation (RAG), JUnit / Mockito, Unit Testing, Scalable Systems, Backend Architecture, Systems Design, Amazon Web Services (AWS), Github Actions, CI/CD, Kubernetes, Docker, SQL, Python, NoSQL Databases, Redis, PostgreSQL, pgvector

Experience

Senior Software Engineer / AI

Verizon

2024 - Present · 2 yrs

Cary, North Carolina, United States

Environment: Java 17/21, Spring Boot 3.3, Spring AI, Python, LangGraph, Pinecone, GitHub Copilot. Multi-Agent Systems: Architected a Multi-Agent Collaboration (MAC) framework utilizing Spring AI and LangGraph to automate complex, multi-step business logic, reducing operational overhead by 40%. Advanced RAG Pipelines: Engineered a sophisticated Retrieval-Augmented Generation (RAG) system using pgvector and semantic search; implemented "Hybrid Search" to ensure high-precision retrieval for internal documentation. AI Guardrails & Safety: Developed a robust Guardrail layer to intercept LLM outputs, ensuring compliance with corporate data policies and preventing hallucinations in customer-facing interactions. Human-in-the-Loop (HITL): Designed and deployed a feedback orchestration module where AI-generated decisions are staged for human approval, utilizing the data to iteratively improve agent prompts. Tool Integration: Created seamless Tool/Function Calling interfaces allowing LLM agents to execute secure Java-based methods for real-time data fetching and transaction processing.

Software Engineer

Bank of America

2022 - 2024 · 2 yrs

Charlotte, North Carolina, United States

Environment: Java 11/17, Spring Boot, Python, AWS, PostgreSQL, Docker, LangChain. AI Transitioning: Led the initiative to integrate AI capabilities into the core Java product suite, implementing GitHub Copilot workflows that improved team coding velocity by 30%. Microservices Evolution: Refactored legacy monolithic services into Spring Boot microservices, significantly improving system modularity and enabling isolated AI feature deployments. Python Integration: Developed Python-based microservices for data preprocessing and embedding generation, integrated via REST into the primary Java ecosystem. API Design: Developed and documented high-performance RESTful APIs to support 1M+ monthly active users, ensuring sub-200ms response times.

Junior Software Engineer (Java)

Southwest Airlines

2020 - 2022 · 2 yrs

Dallas, Texas, United States

Environment: Java 8/11, Spring Framework, MySQL, RabbitMQ, Maven, Jenkins. Backend Development: Owned the development of core business modules in a high-concurrency Java environment, focusing on scalability and clean code principles. Database Optimization: Optimized complex SQL queries and schema designs, reducing database load by 15% during peak traffic hours. Unit Testing: Maintained 90% test coverage using JUnit and Mockito, ensuring high-quality releases in a fast-paced CI/CD environment.