Headline

AI Engineer at Fractal | Multi Agent Production grade systems

About

As a Senior AI Engineer at PwC India, I lead the engineering team on a healthcare regulatory project, focusing on developing LLM-powered multi-agent workflows. My recent contributions include designing and deploying RAG pipelines with hybrid retrieval in OpenSearch and Redis for context caching, as well as implementing a multi-agent orchestration system using a Supervisor-Worker pattern with JSON schema-based toolchains. These efforts enhance modularity and enable efficient automated workflows. A Computer Science graduate from PES University, I bring expertise in multi-channel analytics, knowledge graph-based search, and AWS Lambda to enterprise AI implementations. My current work includes building a real-time evaluation dashboard for GenAI virtual assistants, ensuring proactive monitoring of response quality. Passionate about scalable and innovative AI solutions, I am motivated by opportunities to advance the boundaries of generative AI applications.

Location

Bengaluru

AI stack

hugging face

langchain

langgraph

llama.cpp

LlamaIndex

Ragas

Work experience

Artificial Intelligence Engineer at Fractal, Sep 2026 - Present · 1 mo, Bengaluru

Senior AI Engineer at PwC India, May 2026 - Aug 2026 · 4 mos, Karnataka, India, Lead the engineering team on a healthcare regulatory project, owning architecture and delivery of LLM-powered multi-agent workflows for the last 6 months. Designed and deployed RAG pipelines with hybrid retrieval (BM25 + embeddings) in OpenSearch and Redis for context caching/state management, applying prompt engineering to improve response quality and reduce token usage. Built a multi-agent orchestration system using Supervisor-Worker pattern with dynamic tool calling via JSON schemas in DynamoDB, enabling modular LLM agent capabilities for automated workflows. Implemented a real-time live evaluation dashboard for the multi-agent GenAI virtual assistant, featuring a proactive evaluation node that continuously monitored response accuracy and automatically triggered interventions or user notifications on low-confidence outputs. Developed backend services as gRPC microservices on Amazon EKS with horizontal pod autoscaling and observability, focusing on reliable, low-latency deployment of AI features at scale. Built monitoring and automated incident response using AWS Step Functions and SNS, achieving 99.5% uptime and reducing mean time to recovery by 60%. Previously developed conversational AI solutions including a customer-facing chatbot and a custom medical image classifier using a domain-specific CNN. Implemented custom RAG pipelines entirely in Python using FAISS for vector storage and retrieval orchestration without heavy frameworks.

AI Engineer at PwC India, Jul 2023 - May 2026 · 2 yrs 11 mos, India

Intern at PwC India, Jan 2023 - Jul 2023 · 7 mos, India, Built shipment automation pipelines for a multi-country FMCG client using Azure Functions, Logic Apps, and OCR with Power BI analytics. Enhanced an existing production analytics system by integrating additional data sources and improving processing logic based on client requirements.

Web Developer at Basel Practitioners Pvt. Ltd., Aug 2022 - Dec 2022 · 5 mos, Bengaluru, Built data pipelines from AWS and developed interactive Power BI dashboards for business reporting.

Education

Bachelor of Engineering - BE, Computer Science, PES University

Public links

samhitharao908.github.io: https://samhitharao908.github.io/portfolio/