Eshika Aggarwal
Fremont,CA
Summary
AI and Software Engineer with a robust foundation in Data Science from the University of California, Berkeley. Proven track record in architecting advanced AI orchestration layers, specializing in Model Context Protocol (MCP) and Agent-to-Agent (A2A) communication. Expertise in Python and cloud-native infrastructure enables the development of innovative agents equipped with comprehensive tools to drive efficiency and foster innovation. Committed to leveraging cutting-edge technology to tackle complex challenges and enhance operational capabilities across diverse environments.
Overview
3
3
years of professional experience
Work History
Software Engineer
IBM
San Jose, CA
01.2025 - Current
- Architected a dynamic tooling layer for an AI agent using Model Context Protocol (MCP), allowing agents to autonomously register and consume tools from external MCP services via endpoint/auth registration and available MCP operations.
- Developed a 'Spec-to-Server' tool that converts OpenAPI specifications into deployable MCP servers on IBM Cloud, significantly reducing tool integration cycles.
- Engineered Agent-to-Agent (A2A) communication protocols for an agentic library framework, enabling 10+ downstream IBM AI agents to share context and hand off tasks autonomously.
- Led a registry project to enable MCP tools directly via an agent chat. The same could also be done for agents. Augmented with automated description/example generation for LLM context, improving agent and tool selection accuracy by over 40%
- Modernized product security scanning by migrating to Tekton (IBM Cloud) and implementing parallelization, reducing scan runtime from 24+ hours to a scalable automated process taking just 1 hour.
Software Engineer I
Cisco Systems
07.2024 - 08.2024
- Accelerated critical error/warning isolation by 50% via a real-time topological visualization that mapped spine-leaf hierarchies to pinpoint specific nodes of failure and recommended remediation steps
- Implemented path visibility measures for multicast faults in complex distribution trees visualizing PIM (Protocol Independent Multicast) and IBMP states
- Developed API handlers aligned with OpenAPI Specification (OAS), resulting in a 40% faster data retrieval process for Nexus Dashboard users.
- Refactored and consolidated legacy UI/API code from multiple repositories into a scalable ReactJS codebase, reducing technical debt.
Data Scientist Intern
Cisco Systems
06.2023 - 08.2023
- Developed a connectivity analysis feature to simulate Cisco's 3-tier server topology using Python and Kubernetes for traffic routing visualization.
- Engineered an automated diagnostic tool correlating multicast states with data-plane telemetry to conduct RCA (root cause analysis) on package drops in high-bandwidth streams
Education
B.A. - Data Science
University of California, Berkeley
05.2024
Skills
● AI & Orchestration: Agentic Frameworks, Model Context Protocol (MCP), RAG, A2A Communication, Agent Development
● Languages: Python, TypeScript, JavaScript (React), Shell, SQL, Java, C
Timeline
Software Engineer
IBM
01.2025 - Current
Software Engineer I
Cisco Systems
07.2024 - 08.2024
Data Scientist Intern
Cisco Systems
06.2023 - 08.2023
B.A. - Data Science
University of California, Berkeley
LEADERSHIP
- Women in Computing and Data Science (WiCDS) | Treasurer
- Curated and taught Python/SQL modules for 30+ students; saw a 30% average improvement in coursework grades.