Arnab Kundu

Associate Consultant

Kolkata, West Bengal, India4 yrs 8 mos experience

Key Highlights

Expert in Generative AI and LLM fine-tuning.
Proven track record in deploying scalable AI solutions.
Strong experience in cloud-native architectures and MLOps.

Stackforce AI infers this person is a skilled AI Engineer specializing in cloud-native solutions for Fintech and SaaS industries.

Contact

Skills

Core Skills

Generative AiLlm Fine-tuningCloud DevelopmentMicroservicesDevopsCloud Infrastructure

Other Skills

GenAiPython (Programming Language)LangchainFlaskPythonJavaAmazon EKSSpring BootAzure Kubernetes ServiceMicrosoft AzureAzure Cosmos DBAzure DevOpsTerraformAngularDocker

About

AI Engineer with 4+ years of experience building and deploying enterprise-scale AI solutions across banking, insurance, and compliance domains. Specialized in Generative AI, Agentic AI, Retrieval-Augmented Generation (RAG), LLM fine-tuning, and cloud-native AI platforms. Currently working at EY GDS, where I design and deploy production-grade AI systems using Azure OpenAI, LLaMA-3, NVIDIA NeMo Retriever, LangChain, LangGraph, FastAPI, Kubernetes, Azure, and AWS. I have experience developing large-scale RAG platforms supporting hundreds of enterprise users, indexing tens of thousands of documents, and delivering low-latency, high-throughput AI applications in regulated environments. My expertise includes LLM fine-tuning using QLoRA, multimodal AI, AI agents, workflow orchestration, vector databases, hybrid search, MLOps, CI/CD automation, and scalable cloud architectures. I enjoy solving complex business problems by combining machine learning, software engineering, and distributed systems to deliver measurable impact. Key Areas: • Generative AI & LLMs • Agentic AI & Multi-Agent Systems • RAG & Knowledge Retrieval • LangChain & LangGraph • NVIDIA NeMo & Azure OpenAI • FastAPI, Python & React • Kubernetes, Docker & MLOps • Azure & AWS Cloud Architecture • CI/CD, DevOps & Enterprise AI Platforms

Experience

4 yrs 8 mos

Total Experience

Average Tenure

Current Experience

Ey

Senior Analyst - AI

May 2025 – Present · 1 yr 1 mo · Kolkata, West Bengal, India · Hybrid

Ltimindtree

3 roles

Associate GenAI Developer

Nov 2022 – May 2025 · 2 yrs 6 mos · Kolkata, West Bengal, India · Hybrid

Fine-Tune LLM Model with Domain Data
Fine-tuned large language models using 7,000 actual data points, expanding to 140,000 data points with a Neo4j knowledge graph using Python.
Trained these models on Azure ML Studio and AWS SageMaker, successfully deploying them in Azure Kubernetes Service for inference and implementing an Azure DevOps CI/CD pipeline.
Developed and maintained web applications using Flask and React, enhancing user experience and performance.
In-House Generative AI Python Framework
Developed an in-house Python framework following the Abstract Factory Pattern for structured data extraction from unstructured documents, leading to the successful launch of AI-enabled products using Langchain and Flask.
Integrated an API Gateway with RESTful services for seamless API development using Flask.
Collaborated with cross-functional teams to ensure high-quality, scalable API delivery while adhering to industry best practices.
Insurance-Based AI-Enabled Platform
Developed an insurance-based AI-enabled platform using Python and Langchain, incorporating advanced AI capabilities such as a Q&A chatbot and topic summarizer utilizing an advanced RAG mechanism.
Deployed the platform in Elastic Kubernetes Service (AWS) using the `eksctl` CLI for effective infrastructure management.
Utilized various AWS services, including EC2, S3, Secret Manager, Certificate Manager, ECR and EKS, to optimize cloud-based solutions.

GenAiPython (Programming Language)Generative AILLM Fine-Tuning

Associate Cloud Developer

Promoted

Sep 2022 – Oct 2024 · 2 yrs 1 mo · Kolkata, West Bengal, India · Hybrid

End-to-End Email Workflow Automation and Submission Platform
Developed an email-triggered orchestrator layer using Java Spring Boot, integrating with Spring Data JPA, Lombok, Hibernate, and Microsoft Logic App for seamless workflow automation across microservices. Deployed all microservices in Azure Kubernetes Service.
Designed and implemented microservices for file processing and quote generation, interfacing with Duck Creek Server and displaying results on the HIL Screen using Angular.
Optimized system efficiency by automating end-to-end workflows, utilizing Spring Boot, Spring JPA, Azure Cosmos DB, Azure SQL Server, and Lombok to reduce processing time and enhance modularity.
Insurance-Based AI-Enabled Platform
Developed an insurance-based AI-enabled platform using Java, Spring Boot, Python, and Langchain, incorporating advanced AI capabilities such as a Q&A chatbot and topic summarizer utilizing an advanced RAG mechanism.
Deployed the platform in Elastic Kubernetes Service (AWS) using the `eksctl` CLI for efficient infrastructure management.
Utilized various AWS services, including EC2, S3, Secret Manager, Certificate Manager, ECR, and EKS, to optimize cloud-based solutions.

JavaAmazon EKSCloud DevelopmentMicroservices

Associate DevOps Engineer

Oct 2021 – Mar 2023 · 1 yr 5 mos · Kolkata, West Bengal, India · Hybrid

Duck Creek Automation Project
Developed a Blue-Green Deployment Solution through R&D for a US-based client’s development environment, involving 17 .NET-based microservices deployed in Azure Kubernetes.
Implemented CI/CD for multi-tier architecture microservices using Azure DevOps Pipelines.
Wrote modular Terraform scripts to automate the creation and deletion of inactive environments and manage Blue-Green deployment switches for VM-based setups.
Successfully migrated Octopus Deploy pipelines to Azure DevOps pipelines, streamlining CI/CD processes and integrating automated deployment workflows within the Azure DevOps environment.
Duck Creek Cloud Tagging Project
Developed a platform for bulk tagging of Azure resources using Angular for the frontend and Python for backend functionality.
Configured CI/CD pipelines by writing YAML scripts in the Azure DevOps portal for each stage of deployment, successfully deploying in Azure Container Apps.
Duck Creek Transformation COE Project
Integrated Kubernetes auto-scaling functionality into Azure DevOps agents using KEDA, optimizing resource utilization and scaling based on demand.
Created custom Docker images for both Windows and Linux base environments, optimizing them following best practices, and deployed them into Azure Kubernetes Service using Helm charts.