Jagan P

About Jagan P

I am an NLP Engineer at Zoho specializing in architecting and managing complete Retrieval-Augmented Generation (RAG) pipelines, pre-training & fine-tuning large language models, and designing scalable agent frameworks for production.Over the last 6+ years, I have- Built & owned Zoho’s RAG Chat system, achieving 87% exact-match accuracy on daily queries and sub-second response times- Led an LLM pre-training & fine-tuning team (3 engineers), using model-distillation and prompt-engineering to boost model performance by 30% while cutting inference costs by 25%- Architected & developed the Agent Studio framework powering Zoho’s AI product- Drove product launch, collaborating across PM, UX, and DevOps teams to ship a fully integrated AI assistant featured in Zoho’s Agent Studio.I thrive at the intersection of research and production - whether it’s iterating on novel LLM architectures, automating MLOps pipelines, or mentoring teams on best practices.

Experience

1. Natural Language Processing Engineer

   Zoho

   Aug 2021 — Present · Chennai, IN

   Built & own Zoho’s end-to-end RAG Chat system—training T5-based retriever/reader models that deliver 87% exact-match accuracy on daily queries with sub-second latency- Lead a 2-engineer LLM pre-training & fine-tuning team, leveraging model-distillation and prompt-engineering to boost performance by 30% while cutting inference costs by 25%- Architected & developed the Agent Studio framework powering Zoho’s flagship AI assistant, using modular PyTorch pipelines and Kubernetes auto-scaling to support concurrent sessions- Drove a successful product launch, coordinating across PM, UX, and DevOps to ship a fully integrated AI assistant within Zoho’s Agent Studio- Mentored junior engineers on AI best practices: established model-training pipelines, R&D workflows, automated monitoring, and rigorous data-validation processes.

Education

• KG College of Arts and Science (KGCAS)

  Bachelor of Science, Mathematics