Home
// Staff Engineer · AI Systems Architect

Sumit Kar

Building resilient backend systems and orchestrating AI-native architectures at enterprise scale.

Staff Engineer
@ UKG

Sumit Kar

The Architect

10+
Years Engineering
Building production systems at scale across four industry leaders.
4
Companies · Staff → SWE III
UKG, Zuora, Cisco, BroadSoft — from startup energy to enterprise scale.
AI Integrations Built
LLMs, RAG pipelines, agent systems woven into real production workflows.
0
Undetected Outages Goal
Chaos engineering, Gremlin testing, load testing — resilience by design.

Staff Engineer at UKG, architecting backend systems with NestJS, TypeScript and weaving AI/LLM intelligence into enterprise workflows. Deep expertise in microservices, cloud-native design, and making distributed systems fail gracefully.

Passionate about RAG pipelines, agent systems, and the intersection of backend engineering and AI. Author. Mentor. System thinker.

JavaNestJSLangChainRAGKubernetesKafkaSpring BootLangGraphGremlink6DockerGemini

Tech Stack

// Core Languages
IDLE
Java
Java
SYS_CORE
TypeScript
TypeScript
FRONT_EDGE
Python
Python
INFERENCE
NodeJS
NodeJS
RUNTIME
Go
Go
CONCURRENT
JavaScript
JavaScript
ACTIVE
// AI / LLM Systems
IDLE
LangChain
LangChain
ORCHESTRATOR
LangGraph
LangGraph
AGENT_STATE
RAG
RAG
VECTOR_DB
Gemini
Gemini
MODEL
LLMs
LLMs
INTELLIGENCE
Agents
Agents
AUTONOMOUS
// Frameworks & APIs
IDLE
Spring Boot
Spring Boot
ENTERPRISE
NestJS
NestJS
MICROSERVICES
Angular
Angular
SPA_RENDER
REST APIs
REST APIs
PROTOCOLS
WebRTC
WebRTC
REALTIME
// Infrastructure & DevOps
IDLE
Docker
Docker
CONTAINER
Kubernetes
Kubernetes
ORCHESTRATE
Jenkins
Jenkins
PIPELINE
GitHub Actions
GitHub Actions
CI/CD
Ansible
Ansible
AUTOMATION
Docker Swarm
Docker Swarm
SWARM
// Data & Messaging
IDLE
Postgres
Postgres
RELATIONAL
MongoDB
MongoDB
DOCUMENT
Elasticsearch
Elasticsearch
SEARCH_IDX
Kafka
Kafka
EVENT_BUS
MySQL
MySQL
LEGACY_DS
// Observability & Testing
IDLE
Grafana
Grafana
METRICS
Gremlin
Gremlin
CHAOS_ENG
k6
k6
LOAD_TEST
Gatling
Gatling
LOAD_SIM
JMeter
JMeter
STRESS
Lighthouse
Lighthouse
PERF_AUDIT

AI & Product Thinking

LLM Integration
Embedded large language models into backend workflows using LangChain and Gemini — powering AI-native features at enterprise scale.
RAG Pipelines
Built retrieval-augmented generation systems that deliver context-aware, grounded AI responses — not hallucinations.
Agent Systems
Designed A2A interfaces and Agent UI components for cross-system orchestration via Backstage, using LangGraph.
AI-Powered Automation
Wired AI workflows into CI/CD and operational pipelines — turning intelligence into leverage, not just features.
INH1LLMRAG
INFERENCE ENGINE · ACTIVE

The Journey

Jun 2024 — Present
~2 yrs · Delhi
UKG
Staff Software Engineer
  • Architected backend services with NestJS & TypeScript using SOLID principles and modular design.
  • Integrated LLMs into workflows via LangChain, LangGraph, RAG, and Gemini — enabling AI-powered automation.
  • Built A2A interfaces and Agent UI components for seamless cross-system Backstage integration.
  • Managed Kubernetes deployments for fault-tolerant, high-availability services.
  • Conducted chaos & resilience testing with Gremlin; load testing with k6, Gatling, Lighthouse.
NestJSLangChainRAGKubernetesGremlink6
Jun 2022 — Feb 2024
1yr 9mo · Chennai
Zuora
Software Engineer III
  • Optimized payment workflows, cutting onboarding time for new gateways significantly.
  • Built and stress-tested REST APIs with JMeter for data integrity at scale.
  • Monitored & optimized application performance metrics using Grafana.
  • Championed knowledge sharing through internal tech talks and mentoring.
JavaSpring BootGrafanaJMeterREST
May 2018 — Jun 2022
4yrs 2mo · Chennai
Cisco
Software Engineer
  • Built a Network Testing Tool using Java, JavaScript, and WebRTC for infrastructure optimization.
  • Contributed to microservices in Java, NodeJS, Python & Go across multiple product lines.
  • Orchestrated multi-node Docker Swarm deployments; secured images with Ansible.
  • Developed Python-based call quality threshold alerts for proactive monitoring.
JavaWebRTCDocker SwarmPythonGoAngular
Jul 2016 — May 2018
1yr 11mo · Chennai
BroadSoft
Software Engineer
  • Engineered Java REST APIs for MySQL & Elasticsearch data retrieval.
  • Built a Device Provisioning Portal — improved dashboard load time by 110%.
  • Deployed system components as microservices via Docker for elastic scalability.
  • Validated stability with JUnit, Gatling, and JMeter across performance benchmarks.
JavaElasticsearchDockerGatlingJUnit
Jan 2016 — Apr 2016
4 mos · Kolkata
Novatree eSolutions
Internship
  • Created responsive websites using HTML5, CSS3, JavaScript, PHP (Laravel) and Drupal CMS.
  • Enhanced MongoDB data processing efficiency by 25%.
PHPLaravelDrupalMongoDBHTML5CSS3

Shipped. Scaled.

Enterprise AI Workflows
LLM → Production · UKG
Designed and shipped RAG pipelines and LangGraph-based agent systems that power real AI features in UKG's enterprise platform — not demos, real users.
Payment Gateway Optimization
Onboarding Time ↓ · Zuora
Streamlined payment processing workflows to dramatically cut gateway onboarding time, directly improving tenant experience at Zuora's SaaS platform.
Network Testing Tool
WebRTC + Java · Cisco
Built a comprehensive web-based network infrastructure testing tool with real-time WebRTC capabilities — giving teams visibility into call quality at scale.
Dashboard Performance
+110% Load Speed · BroadSoft
Overhauled a device provisioning portal's frontend performance — achieving 110% improvement in load time through targeted optimization strategies.
K8s Platform Engineering
Zero Downtime Deployments
Managed Kubernetes-based service deployments with CI/CD automation via GitHub Actions — ensuring fault tolerance across all environments.
Backstage A2A Integration
Cross-system Agent UI
Developed A2A interfaces enabling seamless agent-to-agent communication through Backstage — a foundation for AI-native developer portals.

Digital Bharat AI & Computing

Digital Bharat book cover

A practical exploration of AI, computing, and the systems thinking required to build digital infrastructure that scales beyond demos.

The book is available as an interactive flip-book reader on the site, as a downloadable PDF, and in physical edition via Amazon and Flipkart.

Chaos by Design

Systems break in production. The question is: did you design for it? Chaos engineering means engineering for failure — deliberately, methodically, calmly.

99.9%
Uptime Target
4
Testing Tools
0
Panic Incidents
Gremlin
k6
Gatling
JMeter
Lighthouse
"
Systems should fail gracefully.
Intelligence should scale quietly.
— Sumit Kar · Staff Engineer · AI & Backend Systems

Let's
Connect.

Staff Engineer. System thinker. Always open to interesting problems.