Rupendra K

Data Scientist

Bell Canada • Freiberufler

Jan 2023 - Present • 3 yrs 7 mos

Working on the design and deployment of production-grade AI and agentic systems to improve operational efficiency across large-scale enterprise environments. My focus is on building reliable, scalable AI workflows that move beyond experimentation into real-world, business-critical usage. Key contributions include designing and deploying an AI-powered Slack chatbot using Python, Flask, and the Slack Bolt SDK, with a modular, testable architecture supported by CI/CD pipelines in Docker and GitHub Actions. This system significantly reduced false escalations (~40%) and accelerated operational response times (~70%). I have built and maintained RAG pipelines and retrieval systems, implementing hybrid search strategies (dense and sparse retrieval) and integrating transformer-based models to improve context-aware responses while reducing retrieval latency. These systems were deployed on GCP with production monitoring and performance optimization. A core part of my work involves agentic orchestration and enterprise integrations. I designed and managed Model Context Protocol (MCP) servers in Python to securely connect AI agents with tools such as Jira, Confluence, and observability platforms. This enabled governed tool-calling, scalable inference, and seamless workflow automation across internal systems. I’ve also worked on LLM fine-tuning and deployment, adapting open-source models (e.g., Mistral, Mixtral) within agentic frameworks to improve domain-specific accuracy and enable reliable zero-shot and retrieval-augmented behaviors in production. Recognized with Excellence and Innovation Awards for delivering high-impact AI systems that automated decision workflows, reduced manual review effort, and improved operational intelligence for enterprise stakeholders.