Naresh Tinnaluri

Engineered a robust Multi-Agent Framework leveraging OpenAI GPT-4 Turbo and Gemini AI, orchestrated with LangChain & LangGraph, significantly improving system efficiency and complex task execution.

Pioneered the development of Model Context Protocol (MCP) to facilitate seamless Agent-to-Agent (A2A) communication, enhancing AI reasoning, coherence, and task-specific performance.

Successfully deployed multi-agent systems for health tech (nutrition, diagnosis, wellness) under a supervisor-aggregator architecture, improving decision support and streamlining patient care processes.

Optimized LLM performance through fine-tuning with Chain of Thought (CoT) techniques, leading to enhanced contextual reasoning across specialized health domains.

Led scalable deployments on AWS (Bedrock, EC2, Lambda, S3) with Dockerized, fault-tolerant cloud integration, ensuring high availability and performance for critical AI services.

Designed and launched production-grade Python APIs for a multi-agent health AI system (MOA) integrating LLMs like Phi-2 and Meditron-7B, enabling seamless integration with existing platforms.

Developed and scaled RAG pipelines and Chain of Thought (CoT) enhancements, demonstrably reducing AI hallucinations and improving factual accuracy.

Architected decentralized multi-agent systems leveraging MCP and A2A protocols, integrating IPFS and Blockchain to ensure immutable data tracking and enhanced bias detection.

Achieved a 30% reduction in hallucinations and biases within multi-agent pipelines through advanced CoT prompting and chaining techniques.

Designed Directed Acyclic Graph (DAG) structures for real-time agent interaction tracking, significantly improving explainability and mitigating biases in LLM responses.

Automated over 100 form submissions using LangChain agents, A2A communication, and OpenAI GPT-4 Turbo, significantly enhancing web scraping and intelligent form completion capabilities.

Boosted system scalability and performance by optimizing AWS Cloud infrastructure, LangGraph, and agentic RAG pipelines to support increased data processing volumes.

Developed robust backend services with FastAPI, LangGraph, and OpenAI APIs, automating venture capital and fellowship form submissions to streamline application processes.

Achieved a 75% reduction in workflow automation time by seamlessly integrating LangChain agents with Playwright for comprehensive process optimization.

Architected and deployed cloud-first solutions on AWS with Docker, ensuring low-latency, high availability, and optimal performance for automated systems.

Led the development of an Autonomous Torpedo featuring real-time object detection using YOLOv5 and U-Net, deployed on Jetson Nano via TensorRT and ONNX, showcasing advanced edge AI capabilities.

Optimized AI pipelines with Docker, achieving a 70% reduction in computational cost and a 50% boost in model accuracy.

Developed and deployed robust web backends using Django and FastAPI with Docker, supporting critical application functionality.

Engineered modular login/authentication APIs and certificate generation systems, achieving a 25% speedup in throughput.