RAG Ingestion & Agent Orchestration • Integrated AWS Bedrock Claude into the Agent Orchestrator using LiteLLM, designing a multi-cloud LLM abstraction with provider-specific defaults and validating Claude configurations against Azure OpenAI baselines through routing performance analysis and regression testing. • Prototyped pre-inference prompt validation with Bedrock Guardrails and Azure AI Content Safety to detect banned words and sanitize unsafe user inputs before model execution. • Contributed to re-architecting the Retrieval-Augmented Generation (RAG) ingestion system, transitioning a single synchronous service into an event-driven Kafka pipeline of three decoupled microservices (ingestion, chunking, embedding) for reliable large-file processing and independent scaling. • Designed REST APIs to ingest and index multiple document types (PDF, Markdown, plaintext) using Sentence Transformers and OpenAI embedding models. • Built a cloud-agnostic storage handler library that routes documents to Azure Blob Storage or AWS S3 based on the service cloud provider, later serving as the storage layer for the Manifest Configuration Service. • Developed tenant-level configuration and sharing workflows for Brand Concierge using A2A Manifests and Agent Cards, facilitating broader customer adoption. • Enhanced the Manifest Configuration Service with an async blob-to-blob migration tool that upgrades legacy cloud-stored manifests to typed code schemas during service startup. Cloud Infrastructure & CI/CD (AWS + Azure / Kubernetes) • Built a Jenkins PR validation pipeline for A2A manifest changes that publishes manifests to Azure Blob Storage, provisions ephemeral Agent Orchestrator environments, and runs pre-merge regression suites. • Deployed and operated services on AWS and Azure Kubernetes (Argo CD, Helm), overseeing rollouts, end-to-end validations, scaling, monitoring (Prometheus/Grafana), and incident response. • Participated in on-call rotations to diagnose and resolve production incidents across multiple services, maintaining high system reliability. • Mentored engineers and led knowledge-sharing sessions across the team.

Document Ingestion Platform • Led comparative POCs of vector databases (Pinecone, Milvus, pgvector) for storing document embeddings, assessing query performance, scalability, and operational overhead. • Implemented Kafka/PostgreSQL health-check monitoring wired into Kubernetes readiness probes, async end-to-end tests, and document retention/delete policies across Azure Blob Storage and AWS S3. • Built multiple Jenkins pipelines to publish shared Python libraries to internal package registries, with automated version-bump on release.

Roomio: Developed features for a workspace-booking platform covering meeting rooms, desks, parking spaces, shared spaces, and visitor self-registration — backend with Java, Spring Boot, and PostgreSQL, and frontend flows in JavaScript, HTML, and CSS for reservations, availability views, and admin management. Trax: Developed backend business logic and UI improvements for an internal employee and project-management platform using Java, Spring Boot, Spring Security, REST, PostgreSQL, and Redis, adding features that improved visibility into company operations.

Designed and developed an internal project-estimation web application that translated requirement inputs into structured implementation-effort calculations for software delivery planning.

Relevant Coursework: Artificial Intelligence, Multi-Agent Systems, Information Retrieval, Advanced Intelligent Web Systems, Advanced Web Technologies, Architectural Styles and Software Design Patterns, Distributed Internet Processing.

Relevant Coursework: Data Structures and Algorithms, Object-Oriented Programming, Programming Paradigms, Communication Protocols, Algorithm Design, Operating Systems, Parallel and Distributed Algorithms.