Open to opportunities

Anand Kumar

@anandkumar14

Message

I build production-grade generative AI, multi-agent, RAG, and computer vision solutions.

India

Message

What I'm looking for

I'm looking to build scalable, production-ready generative AI, agentic AI, RAG, and computer vision solutions that improve automation, operational efficiency, and business outcomes.

I'm building enterprise AI automation at Q1 Technologies for Publicis Resources, where I architect scalable multi-agent workflows for media-plan extraction, advertising order creation, and document processing using Azure AI Foundry, Azure OpenAI, Python, and Agent@Scale.

I design Planner-Expert-Critic-Saver agent architectures, RAG pipelines, business-rule engines, and LLM evaluation systems that produce validated structured outputs for downstream ERP systems. I also build cloud-native services with Docker and Azure Blob Storage for secure ingestion, scalable processing, and production deployment.

Previously at Logic Fruit Technology, I built multimodal RAG assistants and optimized LLM inference on NVIDIA Jetson and L40S GPUs. I also delivered a YOLOv11 helmet-detection system with 90%+ real-time accuracy and reduced edge latency by 30% using TensorRT.

Experience

Work history, roles, and key accomplishments

Current

AI/ML Engineer

Current

Q1 Technologies India Private Limited

Aug 2025 - Present (1 year)

Architected and deployed scalable Agentic AI workflows using Azure AI Foundry, Agent@Scale, and Azure OpenAI to automate enterprise media-plan extraction and document processing. Designed a Planner-Expert-Critic-Saver multi-agent architecture with MCP integration and RAG pipelines for accurate data extraction.

Azure AI Foundry Azure OpenAI Agent@Scale Python Multi Agent Systems LangGraph RAG Model Context Protocol Docker Azure Blob Storage LLM Evaluation Precision Recall F1 Score

R&D Engineer (AI/ML)

Logic Fruit Technology Pvt Ltd

Jun 2022 - Aug 2025 (3 years 2 months)

Built a multimodal GenAI virtual assistant with RAG and edge AI, optimizing LLM inference on NVIDIA Jetson and L40S GPUs. Designed and deployed a YOLOv11-based helmet detection model with 90%+ accuracy and accelerated edge inference using TensorRT.