Jothiswaran Arumugam — AI Researcher

4+ years building and deploying production-grade AI systems for enterprise clients in financial services and healthcare. I lead technical discovery, architecture scoping, and hands-on delivery for GenAI engagements at RevStar, an AWS Advanced Tier Partner. Over the past year, I've led 25+ client engagements and built: - Agentic AI pipelines using Strands Agents SDK and Bedrock AgentCore for healthcare and fintech clients - A multimodal video-to-brief system with Amazon Nova Pro, processing 3,867 brand videos with 72x cost reduction - AI transaction categorization with Bedrock AgentCore, reducing inference latency from 8.3s to 1.2s and cutting costs by 93% - Multi-agent RAG pipelines on OpenSearch serving 10+ clients with hybrid retrieval - Internal quickstart frameworks that cut delivery timelines by 60% department-wide What I work on day-to-day: - Building agentic solutions using Strands SDK, Bedrock AgentCore, and LangGraph - LLM fine-tuning (QLoRA, distributed training on A100 GPUs) - Multimodal AI with Amazon Nova Pro - Production infrastructure with AWS CDK (TypeScript), CI/CD, and OpenTelemetry observability - AI governance frameworks, prompt versioning, and LLM evaluation (RAGAS, LangSmith) Before RevStar, I built RAG systems and fine-tuned LLMs at IM Digital, and spent 2.5 years at Cognizant building ML infrastructure on GCP. I hold 13 cloud certifications - 6x AWS (including GenAI Developer Professional and Solutions Architect Professional) and 7x GCP (including ML Engineer and Cloud Architect). I also write Jo's Cloud AI Hub, covering practical cloud and AI engineering.

Stackforce AI infers this person is a Data & AI Engineer with expertise in Fintech and Healthcare sectors.

Location: Tallahassee, Florida, United States

Experience: 3 yrs