Divyam Agrawal

About Divyam Agrawal

I’m a Data Scientist focused on building production-grade AI and data automation systems—especially agentic workflows, RAG chatbots, and scalable pipelines. I take ideas from ambiguity to impact by combining problem framing, stakeholder alignment, system design, implementation, UAT, and continuous iteration into reliable, user-ready solutions.I enjoy working at the intersection of applied AI and analytics: orchestrating LLM workflows, retrieval systems, and automation frameworks that turn fragmented knowledge and messy operational data into clean, trustworthy outputs. My work typically centers on improving relevance and response quality, strengthening validation and reliability, and enabling teams to move faster through end-to-end automation.I’m motivated by building AI products that actually ship—systems that are scalable, maintainable, and measurable—using techniques like RAG, vector search, agentic reasoning, prompt engineering, evaluation/QA workflows, and robust data pipelines.

Experience

1. Data Scientist i

   Cvent

   Jun 2024 — Present

   Built and shipped multiple AI chatbots (RAG + agentic) using vector search and multi-stage LLM workflows, improving knowledge/article matching from ~50% to ~85–90% and reducing review/QA effort significantly.* Created an agentic “reasoning engine” service for article drafting using an 8-step validate → draft → style/fact critique+fix → finalize pipeline, improving output consistency and cutting human iteration cycles.* Delivered large-scale enrichment and standardization pipelines powering a client-facing hospitality product feature, processing 300K+ datapoints and enriching 2M+ records with >95% accuracy, and sustaining automated ingestion of 1K+ datapoints/day.* Built an automated complexity classification/scoring system achieving >95% accuracy and handling new items/day, designed for very large-scale processing.* Hardened AI scoring + validation workflows to reach >95% accuracy across 350K+ processed records and reduced manual effort by >90%.* Automated reporting outputs end-to-end (executive metrics, benchmarking, and deck generation), delivering weekly-refresh, leadership-ready metrics and producing 30+ slides while driving >95% reduction in manual slide-building effort.* Consolidated churn insights across multiple systems and improved the dashboard experience, increasing usability and stakeholder confidence in churn drivers and outcomes.* Delivered rapid frontend architecture for an internal AI assistant and maintained a fast-turnaround enhancement stream, enabling earlier testing/integration and sustained stability through continuous fixes.

Education

• Modern Vidya Niketan Sector 17 Faridabad

  11th - 12th

• Modern Vidya Niketan Sector 17 Faridabad

  9th - 10th

  2016 — 2020

• Punjab Engineering College

  B.Tech, Computer Science and Engineering

  2020 — 2024