Looking for a job

Ameya Murlidhar Karpe

@ameyamurlidharkarpe

AI engineer at Heartland Community Network designing sourced answers from loan guidelines; built TenderGenie, cutting review time by 10+ hours per document.

United States

Message

At Heartland Community Network, I lead the design of a system that gives loan officers instant, sourced answers from guideline documents. I also partner with clients to turn business needs into software, data, and automation solutions.

At DataSmith AI, I built and shipped TenderGenie, an LLM-powered tender-analysis platform that cut review time by 10+ hours per document and helped the startup land a Microsoft for Startups Founders Hub spot. At Kelley School of Business, Indiana University Bloomington, I co-authored peer-reviewed research showing a 15.03% engagement lift and engineered user-isolated similarity features to prevent data leakage.

Experience

Work history, roles, and key accomplishments

Current

Senior Consultant

Current

Heartland Community Network

Jun 2026 - Present (4 months)

Leads design of a system providing loan officers instant, sourced answers from guideline documents. Partners with clients to deliver technical, data, and automation solutions across software development and digital transformation.

LLM RAG Consulting Software Development Data Analysis Automation Dashboarding

Graduate Data Science Assistant

Kelley School of Business, Indiana University Bloomington

Jan 2026 - May 2026 (4 months)

Co-authored peer-reviewed research proving goal-oriented labels lift engagement by 15.03% using an SBERT and cosine-similarity clustering pipeline. Engineered user-isolated similarity features to eliminate data leakage and built a production-grade scraping pipeline for Amazon and Walmart.

Data Scientist

DataSmith AI

Jun 2023 - Jul 2024 (1 year 1 month)

Built and shipped TenderGenie, an LLM-powered RFP analysis SaaS, cutting review time by 10+ hours per document and helping land a Microsoft for Startups Founders Hub spot. Benchmarked LLMs and fine-tuned with LoRA/QLoRA, and architected secure Flask APIs with OCR and document-level context isolation.