// AI Automation & Agentic Systems Engineer

Kaverappa MM

Building systems that think, decide & act autonomously

I design Agentic AI, multi-agent pipelines, LLM-powered automation, and RAG architectures that connect AI to real engineering workflows — from Jira ticket routing and CI/CD orchestration to Graph RAG knowledge systems.

About Me

Engineer at the intersection of AI systems and real-world automation

I'm an AI Systems Engineer based out of India, currently building multi-agent automation at EPAM Systems. My work sits at the edge of LLM orchestration and practical engineering — I care deeply about systems that don't just generate text, but actually do things.

My journey started in full-stack engineering, moved through GenAI product development at Infosys, and has now evolved into designing agentic workflows with Human-in-the-Loop controls, Temporal orchestration, and LLM-driven decision engines.

I believe the most interesting AI work happens when intelligent systems interact with messy real-world infrastructure — Jira tickets, Jenkins pipelines, CI/CD systems, and production databases. That's where I operate.

Outside of work, I explore Graph RAG architectures, build open-source AI projects, and experiment with local models using Ollama and LangGraph.

🤖
Agentic Systems
Multi-agent pipelines with Temporal orchestration, HITL validation, and LLM-driven routing
🧠
LLM Integration
RAG, Graph RAG, embedding pipelines, LiteLLM, LangChain, and fine-tuning with LoRA/PEFT
⚙️
Full-Stack Backbone
Python FastAPI backends, Angular/React frontends, deployed via Docker + Kubernetes on AWS
📍
Based out of India
Open to remote roles and hybrid opportunities globally

Work Experience

From full-stack engineering to LLM-powered agentic systems

System Engineer — AI Automation & Agentic Systems
EPAM Systems · Bengaluru(Remote)
Oct 2025 – Present
  • Built a multi-agent AI system to process Jira tickets and dynamically route workflows across specialized agents handling environment detection, intent analysis, and decision execution
  • Developed LLM-driven pipelines using LiteLLM (Gemini models) to interpret ticket context and trigger automated workflows
  • Designed workflow orchestration using Temporal with signal-based execution and event-driven control
  • Integrated Jenkins for CI/CD automation including pipeline triggering, log extraction, and failure reporting
  • Implemented Human-in-the-Loop (HITL) validation to ensure safe execution of production workflows
  • Built a log analysis pipeline that parses Jenkins logs, stores artifacts in GCS, and posts structured feedback to Jira
  • Designed agent routing logic where outputs from one agent dynamically determine the next execution path
↓ 40% manual intervention Multi-agent routing Temporal orchestration HITL validation
Senior System Engineer — GenAI & Full Stack
Infosys Limited · Mysore
Apr 2022 – Oct 2025
  • Developed a Live Q&A platform supporting 10,000+ concurrent users, boosting real-time engagement by 90%
  • Built an AI-powered chatbot using LangChain, Ollama, and LLMs with RAG architecture — achieving 95% accuracy on document-based Q&A
  • Integrated Stable Diffusion API for text-to-image generation of creative assets within the product
  • Designed PostgreSQL + Redis caching strategy to optimize backend response times under high load
  • Built multistage Docker images and CI/CD pipelines with GitHub Actions, reducing release cycles by 50%
  • Managed Agile sprints, Jira tracking, and PR-based code reviews with cross-functional teams
↑ 90% engagement 95% RAG accuracy ↓ 50% release cycles 10K+ users
System Administrator
Diya Systems Pvt Ltd · Mysore
Dec 2021 – Apr 2022
  • Provided web advisory services, resolved DNS configuration issues, and optimized system performance
  • Configured and troubleshot Linux systems, ensuring infrastructure stability and security hardening
  • Enhanced client solutions through technical communication and cross-team collaboration

Skills & Tech Stack

From LLM orchestration to cloud deployments

🤖
Agentic Systems
LangChainLangGraphTemporalHITLMCPn8nAgent routing
🧠
LLMs & GenAI
LiteLLMOllamaOpenAIGeminiVertex AIHuggingFaceBedrock
🗄️
RAG & Vector
RAG pipelinesGraph RAGFAISSNeo4jEmbeddingsPrompt Engineering
🔬
AI/ML
LoRAPEFTTensorFlowScikit-learnPandasNumPyGAN
⚙️
Backend
PythonFastAPIFlaskNode.jsExpressPostgreSQLRedisMongoDB
🖥️
Frontend
ReactAngularTypeScriptRxJSNgRxHTML/CSS
☁️
Cloud & DevOps
AWS EC2LambdaS3SageMakerDockerKubernetesGitHub ActionsGCS
🔧
Integrations
JenkinsJiraGitHubStable DiffusionspaCyText2SQL

Projects

Personal experiments at the frontier of AI, agents, and full-stack engineering

LangGraph + MCP Multi-Agent System
LangGraph MCP LangChain Llama 3.2 Node.js

Multi-agent system powered by LangGraph and a custom MCP server that routes user requests to specialized agents based on intent. Runs locally using Llama 3.2 through Ollama, demonstrating production-style agent orchestration, intelligent tool routing, and context-aware decision making without relying on cloud-hosted LLMs.

GenAI Text Generator
React Node.js GPT-2 HuggingFace MongoDB

Full-stack Generative AI application using GPT-2 models hosted through Hugging Face for context-aware text generation. Supports customizable prompts, tone and length controls, with a React frontend and Node.js backend managing model inference, API integration, and generation workflows.

Education

Academic foundation in engineering

Bachelor of Engineering
Computer Science / Engineering
Aug 2017 – Oct 2021 CGPA 7.62
Pre-University (PUC)
Science Stream
Jun 2015 – Mar 2017 80.33%
SSLC (10th Grade)
Karnataka Board
Mar 2015 90.56%

Certifications

Verified expertise across AI, cloud, and graph platforms

Claude Certified Architect
NVIDIA — Building LLM Apps with Prompt Engineering
Databricks — Generative AI Fundamentals
Salesforce Certified AI Associate
Neo4j Certified Professional
AWS Cloud Quest: Cloud Practitioner
AWS Technology Architecting
Hugging Face — Agents Course

Open to opportunities

I'm actively looking for roles in AI systems, agentic workflow engineering, and LLM-powered automation. Let's build something that actually thinks.