BACKEND AI ENGINEER · PHOENIX, ARIZONA

HARIPRASATH

Murugasundaram

Backend AI engineer with 4+ years designing and scaling distributed systems, microservice architectures, and high-throughput APIs, serving 5M+ users while cutting latency by 60% and saving $40K/month in infrastructure costs. Builds and ships production AI systems too, from LLM agent pipelines to RAG workflows and HIPAA-compliant clinical tools for healthcare. Contributor to Pion and Google's Agent Development Kit. Excited to solve hard technical problems using modern stacks and comfortable working across cross-functional teams to deliver end-to-end solutions.

LINKEDIN GITHUB LEETCODE

YEARS OF EXPERIENCE IN SCALABLE APPLICATIONS

YEARS · AI & ML SYSTEMS

6M+

USERS SERVED

60%

P95 LATENCY REDUCTION

$40K+

MONTHLY INFRA SAVINGS

10+

PRODUCTION APPLICATIONS

EXPERIENCE

THREE ROLES.
PRODUCTION IMPACT.

FOUR YEARS BUILDING SCALABLE BACKENDS, MICROSERVICES, AND AI SYSTEMS ACROSS HEALTHCARE, CONSUMER APPS, AND ENTERPRISE CLOUD MIGRATIONS.

UNIVERSITY OF ARIZONA

SENIOR APPLICATION DEVELOPER

FEB 2026 – PRESENT · PHOENIX, AZ

Core engineer building HIPAA-compliant AI tooling for clinical staff
AI-powered clinical workflows with React Native, Next.js, and FastAPI
AI agents subsystem that cut workflow creation effort by 40%
Secure HIPAA & FHIR data pipelines for clinical data exchange
Offline-first architecture with secure local storage and background sync

EVENTBEEP

SENIOR BACKEND DEVELOPER

FEB 2022 – AUG 2024 · REMOTE

Backend services for 5M+ users on Node.js and AWS
Cut P95 API latency 60% via Redis caching and Lambda migration
In-house WebSocket + Redis chat — saved $40K+/month in vendor costs
LLM recommendation engine lifted engagement ~30%
Decomposed monolith into microservices on AWS EKS
Fixed caching bug in Pion WebRTC library; shipped REST API for 50+ institutions

IBM

ASSOCIATE DEVELOPER

SEP 2020 – FEB 2022 · REMOTE

Migrated on-prem production systems to IBM Cloud
Database tuning via indexing, materialized views, and query plans
MERN portal on AWS for cloud migration tracking — 40% less manual reporting
Hosted 5+ concurrent production systems on custom Linux stack
Automated provisioning with Terraform across release pipelines

PROJECTS

SHIPPED SYSTEMS.
REAL OUTCOMES.

PRODUCTION AI APPLICATIONS SPANNING RAG PIPELINES, AGENTIC HEALTH TOOLS, AND AUTOMATED VOICE ANALYTICS — FROM PROTOTYPE TO DEPLOYMENT.

2024CODED FOR HUMANS
SHORT VIDEO ANALYSIS
LLM + RAG pipeline to analyze short-form video and generate actionable suggestions. Cut infra costs 90% with serverless GPU deployment. FFmpeg, OCR, and vision models for multi-modal analysis. Full-stack portal for results and access control.
2025U OF ARIZONA
AZX HEALTH APP
React Native health app powered by Google ADK Agentic AI with geolocation, weather, and search tools. Conversational AI extracts structured clinical data for personalized recommendations. Clinical trial in Phoenix with 30+ users; Health Connect SDK for wearable sync.
2025VOICE AI
CALL INTELLIGENCE PLATFORM
End-to-end FFmpeg + LLM pipeline extracting sentiment, action items, and quality ratings from support calls. Reduced manual analysis effort by 80%. Next.js dashboard with filterable analytics backed by MongoDB.
OSSPION
OPEN SOURCE
WEBRTC LIBRARY
Identified and fixed a caching bug in Pion, a core WebRTC open-source library widely adopted across real-time communication products.
OSSGOOGLE
OPEN SOURCE
AGENT DEVELOPMENT KIT
Contributed to the TypeScript SDK for Google's Agent Development Kit during AzX Health app development — agentic AI tooling for production clinical workflows.