Skip to content

Harsha Kumar · Senior Software EngineerDallas, TX

Building resilient distributed systems and intelligent software.

Senior Software Engineer with 12 years of experience designing cloud-native platforms, high-performance APIs, distributed systems, and AI-powered applications using .NET, Azure, AWS and modern LLM technologies.

  • C# / .NET 8
  • Azure & AWS
  • Distributed systems
  • LLMs · RAG · Agents
View My WorkDownload Resume
GitHubLinkedIn
request path
User / App
web · mobile
API Gateway
auth · routing · throttling
Service
svc-a
Service
svc-b
Service
svc-c
Cache
redis
Messaging
queue · topic
Database
sql · nosql
AI / Agents
llm · rag · tools

Engineering Impact

Measured results from production systems.

Reliability and performance outcomes from platform work at Verizon, plus twelve years of delivery.

Years engineering experience
12+
Telecom, banking, financial services, performance engineering and enterprise platforms.
Fewer production disruptions
50%
Retry, circuit-breaker, timeout and fallback policies with Polly and the .NET 8 resilience engine.
Lower database load
60%
In-memory and Redis distributed caching in front of the database.
Faster API responses
40%
Under peak load conditions, delivered by the same distributed caching layer.
Faster integration layer
25%
A high-performance wrapper library around downstream BCS services.

Featured Experience

Verizon Home Platform

Senior Software Engineer · Diaspark (client: Verizon) · 2022 – PresentI lead backend development for the services behind the Verizon Home App, the product Verizon home internet customers use to see and control their router and home network. The work spans API design, resilience, caching, third-party integrations and cloud deployment, along with mentoring engineers, driving architectural decisions and reviewing code.

  1. 01Problem

    Customers expect router controls to respond immediately, but each request depends on downstream router-management (BCS) and security (Bitdefender) services the platform does not own. Those dependencies, together with peak-hour traffic and database pressure, are where latency and failures come from.

  2. 02Architecture

    .NET microservices expose RESTful APIs for each capability of the home network, fronted by Azure API Management and deployed to Azure App Service.

    • Router status and connected devices
    • Wi-Fi configuration and parental controls
    • QoS and Internet Backup
    • Bitdefender and BCS integrations behind dedicated service boundaries

  3. 03Engineering

    Third-party REST integrations are wrapped so the rest of the platform depends on stable, testable contracts rather than vendor APIs. I am the go-to engineer for complex .NET issues and hold the bar on scalability, performance and secure coding in code review.

  4. 04Reliability

    Outbound calls run through explicit resilience policies: retry for transient faults, timeouts to bound latency, circuit breakers to stop cascading failures and fallbacks to degrade gracefully. Application Insights provides diagnostics and operational visibility.

  5. 05Performance

    A two-tier cache (in-memory plus Redis) absorbs repeated reads before they reach the database, and a purpose-built client library reduces the overhead of BCS calls.

  6. 06Impact

    Fewer production disruptions, less load on the database, and faster responses when traffic peaks.

    • 50% reduction in production service disruptions
    • 60% reduction in database load
    • 40% faster API responses under peak load
    • 25% faster responses through the BCS integration library
verizon home · backend
Verizon Home App
Angular 18
Azure API Management
gateway · policies
.NET 6/8 microservicesAzure App Service
  • Router status
  • Connected devices
  • Wi-Fi settings
  • Parental controls
  • QoS
  • Internet Backup
Polly / .NET 8 resilienceretrycircuit breakertimeoutfallback
Cache
in-memory + Redis
Data
MongoDB · SQL Server
Integrations
BCS · Bitdefender
Key Vault
secrets
Application Insights
telemetry
Simplified, illustrative view of the platform's main building blocks.
  • .NET 6/8
  • ASP.NET Core
  • Angular 18
  • Azure
  • AWS
  • MongoDB
  • SQL Server
  • Redis
  • Polly
  • Jenkins

Engineering highlights

  • Resilience architecture

    Retry, circuit breaker, timeout and fallback strategies implemented with Polly and the .NET 8 resilience engine.

    Result
    50%
    fewer production service disruptions
    • Polly
    • .NET 8 Resilience
    • Fault tolerance

  • Caching architecture

    A distributed caching layer combining in-memory and Redis caches in front of the database.

    Result
    60%
    lower database load
    Result
    40%
    faster API responses at peak
    • Redis
    • In-memory cache
    • Distributed caching

  • Integration performance

    A high-performance wrapper library for interfacing with BCS router-management services.

    Result
    25%
    faster API response time
    • C#
    • Library design
    • BCS

  • Cloud architecture

    Cloud-native APIs designed and deployed on Azure, with secrets in Key Vault and telemetry in Application Insights.

    • Azure App Service
    • Azure API Management
    • Key Vault
    • Application Insights

Engineering Expertise

Depth across the stack that keeps production running.

Grouped by the problems they solve rather than listed as keywords.
  • 01

    Backend Engineering

    Production APIs and services in C# and .NET, structured to stay clear as they grow.

    • .NET 8
    • ASP.NET Core
    • C#
    • REST APIs
    • Entity Framework Core
    • LINQ
    • CQRS
    • Clean Architecture
  • 02

    Distributed Systems

    Services that cooperate over networks, queues and caches, and keep working when parts of them don't.

    • Microservices
    • Distributed Caching
    • Event-Driven Architecture
    • Fault Tolerance
    • Async Processing
    • Messaging
    • Kafka
    • SQS
    • Service Bus
  • 03

    Cloud Engineering

    Shipping and running workloads on Azure and AWS with repeatable, automated delivery.

    • Azure
    • AWS
    • Docker
    • Kubernetes
    • Helm
    • CI/CD
  • 04

    Reliability & Performance

    Resilience by design, and optimization driven by telemetry rather than guesswork.

    • Polly
    • .NET Resilience Engine
    • OpenTelemetry
    • Grafana
    • Application Insights
    • AppDynamics
    • Performance Engineering
  • 05

    AI Engineering

    LLM applications, retrieval and agent orchestration, built with the same production discipline.

    • Azure OpenAI
    • OpenAI
    • Claude
    • Microsoft Agent Framework
    • Semantic Kernel
    • LangChain
    • RAG
    • Vector Search
    • Multi-Agent Systems
    • AI Foundry
Languages
C# · TypeScript · JavaScript · Python · T-SQL
Frontend
Angular · Vue.js · Ionic
Data
SQL Server · PostgreSQL · MongoDB · Redis · Elasticsearch

AI Engineering

Building the next generation of agentic software.

I combine traditional software engineering with modern AI architecture: LLM applications, retrieval-augmented generation and multi-agent workflows, held to the same standards of reliability, observability and security as any production service. Agents are only as dependable as the systems around them.
agentic flow
  1. User
    Request or task
  2. AI Application
    API, auth, guardrails
  3. Agent / Orchestrator
    Plans, routes, delegates
  4. Tools / MCP
    Typed, permissioned actions
  5. Enterprise Systems
    Services, data, workflows
Agents act through bounded, permissioned tools rather than direct system access.
retrieval flow
  1. Knowledge Sources
    Docs, tickets, runbooks
  2. Embeddings
    Chunk and encode
  3. Vector Database
    Similarity search
  4. RAG
    Retrieve and ground
  5. LLM
    Answer with context
Retrieval grounds model output in an organization's own knowledge.

How I build with AI

  • Agents inside real architecture

    Models get bounded tools, explicit contracts and the same resilience, observability and security controls as any other service.

  • Grounded, not guessed

    Retrieval over an organization's own knowledge keeps answers specific and traceable to their sources.

  • Humans stay in the loop

    Agents investigate and recommend; consequential actions wait for a person to approve them.

Technologies

  • Microsoft Agent Framework
  • Microsoft Foundry
  • Azure OpenAI
  • OpenAI API
  • Claude API
  • Semantic Kernel
  • LangChain
  • RAG
  • Vector Search
  • Multi-Agent Workflows
  • MCP concepts

AI projects

More AI engineering work will be added here as it is published.

Featured Projects

Selected case studies.

Professional platform work and personal AI engineering, labelled clearly as each.
01 · Distributed Systems / Cloud

Verizon Home Platform

ProfessionalDiaspark · Client: Verizon

Backend platform for the Verizon Home App: large-scale .NET APIs for router management, built for resilience and performance under peak load.

  • Router management, connected devices, Wi-Fi, parental controls, QoS and Internet Backup APIs
  • Polly and .NET 8 resilience policies: 50% fewer production disruptions
  • In-memory and Redis caching: 60% lower database load, 40% faster at peak
  • .NET 8
  • Angular
  • Azure
  • AWS
  • Redis
  • MongoDB
  • SQL Server
Read the case study
read path · resilience + caching
API request
router · devices · Wi-Fi
Resilience pipeline
Polly · .NET 8 resilience
  • Retry
  • Circuit breaker
  • Timeout
  • Fallback
L1 cache
in-memory
L2 cache
Redis
Database
MongoDB · SQL Server
50%
fewer disruptions
60%
lower DB load
40%
faster at peak
Illustrative request path: outbound calls run through resilience policies; reads are served from cache before the database.
02 · Agentic AI / Developer Productivity

AI Incident Triage Platform

PersonalAI engineering project, not production work

An intelligent incident investigation system that coordinates AI agents across observability, documentation and issue-management systems, then hands a root-cause hypothesis and recommended actions to a human for approval.

  • Orchestrator agent delegates to observability, knowledge, code/repository and Jira agents
  • Retrieval over runbooks and documentation grounds each hypothesis
  • Recommended actions require explicit human approval
  • .NET
  • Python
  • Microsoft Agent Framework
  • Semantic Kernel / LangChain
  • LLMs
  • RAG
  • Vector Database
  • MCP
incident triage · agents
Incident
alert · ticket
Orchestrator Agent
plans · delegates · merges
  • Observability Agent
    logs · metrics · traces
  • Knowledge Agent
    runbooks · docs (RAG)
  • Code / Repository Agent
    recent changes
  • Jira Agent
    related issues
Root Cause Hypothesis
Recommended Actions
Human Approval
nothing runs without sign-off
03 · Mobile / Full Stack

Bill Scanner

ProfessionalEurofins

Cross-platform receipt digitization app built with Ionic and Angular, with offline SQLite storage and native camera and file integrations. Shipped to both the Apple App Store and Google Play Store.

  • Led requirements, technical analysis and architecture
  • Offline-first storage with SQLite; camera and file access through Capacitor plugins
  • Automated build and release with GitHub Actions and Ionic Appflow
  • Ionic
  • Angular
  • SQLite
  • Capacitor
  • GitHub Actions
  • Ionic Appflow
App Store(opens in a new tab)Google Play(opens in a new tab)
bill scanner · app + delivery
Camera
Capacitor
File access
Capacitor
Ionic + Angular app
cross-platform UI
SQLite
offline-first storage
GitHub Actions + Ionic Appflow
automated build & release
App Store
iOS
Google Play
Android

Career Journey

Twelve years, six companies, one throughline.

From enterprise .NET applications to distributed cloud platforms, performance engineering and AI.

  1. Apr 2022 – Present

    Diaspark · Client: Verizon

    Senior Software Engineer

    Backend lead for the .NET microservices behind the Verizon Home App: router management APIs, resilience, caching and Azure deployment.

    .NET 6/8 · Angular 18 · Azure · AWS · MongoDB · SQL Server

  2. Sep 2021 – Apr 2022

    Wells Fargo

    Senior Software Engineer

    Full-stack modules for an internal ad campaign management platform, including campaign scheduling, targeting and secure asset storage.

    .NET 5 · Angular 12 · SQL Server · Azure

  3. Feb 2020 – Sep 2021

    Eurofins

    Technology Specialist – Performance Engineering

    Diagnosed memory leaks, CPU spikes and thread contention across .NET applications and delivered RCA and tuning plans, improving response times by up to 50%. Led the Bill Scanner mobile app.

    .NET · AppDynamics · dotTrace · SQL Profiler · Ionic

  4. Aug 2018 – Jan 2020

    Ernst & Young

    Senior Software Engineer

    Workflow automation platform for tax and audit teams, including a reusable Exchange Web Services library and stakeholder dashboards.

    .NET Core · Angular · SQL Server · Azure

  5. Jan 2017 – Aug 2018

    Arcadix

    Software Engineer

    School management and assessment platform with a learning content management system for building tests.

    .NET Core · Angular · SQL Server

  6. Dec 2014 – Dec 2016

    Intellileap

    Software Developer

    Business process management system with a drag-and-drop workflow designer, rule engine and third-party integrations.

    ASP.NET · Web API · SQL Server · WinForms

System Design

How I approach engineering.

  1. Principle 01

    Design for failure

    Distributed systems fail. Build retry, timeout, circuit-breaker and fallback strategies intentionally.

  2. Principle 02

    Measure before optimizing

    Use telemetry, profiling and observability to find actual bottlenecks.

  3. Principle 03

    Build for scale

    Design APIs, caching, messaging and data-access patterns with future load in mind.

  4. Principle 04

    Keep systems maintainable

    Prefer clear boundaries, clean architecture, reusable components and pragmatic engineering.

About

Production engineering, now meeting AI.

I've spent the past 12 years building software across telecommunications, banking, financial services, performance engineering and enterprise platforms.

My background started with .NET application development and evolved into distributed systems, cloud architecture, performance engineering and technical leadership: leading backend development, mentoring engineers, driving architectural decisions and setting the bar in code review.

Today I'm particularly interested in the intersection of traditional distributed systems and AI: building software where LLMs and agents operate safely within well-designed production architectures.

Experience
12 years
Based in
Dallas, TX
Industries
Telecom, banking, financial services, enterprise
Education
Bachelor of Engineering in Computer Science, Visvesvaraya Technological University

Contact

Let's build something that scales.

Interested in discussing Senior, Lead or Staff Software Engineering opportunities involving distributed systems, cloud platforms or AI engineering.

Email MeLinkedInGitHubDownload Resume
Location
Dallas, TX · Open to relocation within the United States