I am an AI Engineer at OptCulture specializing in building production Agentic AI systems and multi-agent copilots. At OptCulture, I architected Xio, an AI Copilot that automates campaign generation, customer segmentation (K-Means/ClickHouse), and RAG workflows with Qdrant. I also build secure Model Context Protocol (FastMCP) servers with OAuth 2.1 PKCE and deterministic SQL compilers for LLM tools. My focus is on making AI systems reliable and cost-effective using Logfire for agent tracing, LiteLLM for dynamic routing, and automated evaluation frameworks. Passionate about action-driven AI, I actively build in hackathons like Google's Agentic Premiere League and contribute to the agentic AI community.
AI-Engineer
OptCulture
- AI Engineer at OptCulture, specializing in designing and deploying production-grade Agentic AI systems, multi-agent architectures, and intelligent automation solutions.
- Developed XIO, an enterprise AI Copilot, powered by a multi-agent architecture of 30 specialized AI agents using the OpenAI Agents SDK, FastAPI, and LiteLLM. Built intelligent triage routing to automate campaign assistance, audience segmentation, customer analytics, content generation, and reporting.
- Reduced LLM token consumption by 70% (10K to 3K tokens per request) through prompt optimization, intelligent model selection, and token usage monitoring with Pydantic Logfire, improving the cost efficiency of AI workflows.
- Engineered a secure Model Context Protocol (MCP) server using FastMCP, OAuth 2.0 with PKCE, and JWT authentication. Developed 30+ AI-accessible tools that enable assistants like Claude and ChatGPT to securely query customer, campaign, product, and retention analytics through ClickHouse.
- Implemented Retrieval-Augmented Generation (RAG) using Qdrant, FastEmbed, and semantic retrieval, alongside real-time Server-Sent Events (SSE) streaming and structured UI responses for interactive AI experiences.
- Built an end-to-end ML-powered customer segmentation platform using K-Means, Non-negative Matrix Factorization (NMF), and Scikit-learn to identify behavioral patterns and purchasing preferences. Containerized and deployed the pipeline using Docker, AWS ECS Fargate, ECR, and S3.
- Passionate about building reliable, scalable, and action-oriented AI systems. Continuously exploring agent orchestration, MCP, LLM evaluation, and emerging AI technologies, with participation in Google's Agentic Premier League and AI engineering initiatives.
An Agentic RAG engine exposed via FastMCP that uses Python, Qdrant, BGE embeddings, Cross-Encoder reranking, Gemini LLM, and Bright Data for self-verifying vector retrieval and selective web search fallback.
Btech in Computer Science
Keshav Memorial institute of Technology · 8.4
Intermeadiate (MPC)
Sri Chaitanya institute of Technology
Secondary School Certificate (ICSE — Class X)
Johnson Grammar School ICSE
Anthropic
Attendee · Oct 24, 2026