Open to opportunities

Riya Shetty

@riyashetty

Message

Data scientist building multi-agent AI and predictive models to optimize marketing workflows and business decisions.

United States

Message

What I'm looking for

I’m looking for a role where I can build production-grade ML/LLM systems—multi-agent workflows, retrieval pipelines, and predictive models—focused on measurable business impact, collaboration across product/marketing, and strong engineering practices.

I’m a Data Scientist focused on turning messy business problems into measurable outcomes. At Smithville Fiber, I build practical AI systems—especially multi-agent LLM workflows—that automate analytics and speed up decision-making across teams.

I’ve delivered a LangGraph-based multi-agent system for Marketing & HR that automates fiber analytics and document-aware Q&A (92% output fidelity), and an MCP server that improves metadata-driven retrieval precision by 18% while reducing latency by 20%. I also fine-tuned a customer churn model (Recall 0.87, F1-score 0.84) and helped deploy an ML-based prioritization framework in Salesforce to boost targeted retention efforts by 11%.

I bring production-minded execution: interpretable ETS forecasting (12% planning efficiency improvement), analytics that raised email conversions by 15% and social engagement by 34%, and a Salesforce migration covering 150K+ case records with performance gains (45% reduction in data retrieval time). Earlier, I owned an internship recommender system end-to-end with Flask + a React frontend—my way of staying curious, rigorous, and impact-driven.

Experience

Work history, roles, and key accomplishments

Current

Data Scientist, Marketing

Current

Smithville Fiber

Jul 2023 - Present (2 years 10 months)

Built a LangGraph-based multi-agent system for Marketing & HR to automate fiber analytics and document-aware Q&A with 92% output fidelity. Developed an MCP server with LlamaIndex + PGVector retrieval (18% precision lift, 20% latency reduction) and implemented a churn model (Recall 0.87, F1 0.84) and Salesforce deployment to boost targeted retention by 11% while optimizing CRM workflows and migrati

LangGraph LlamaIndex Pgvector SQLAlchemy MCP Servers Churn Modeling Salesforce CRM Apex SOQL

Data Science Intern

AI Camp

May 2022 - Aug 2022 (3 months)

Built an internship recommendation system by dynamically scraping 1,000+ job descriptions and 250+ student profiles and training a Word2Vec recommender with a ≥65% cosine similarity threshold. Validated performance via A/B testing (F1: 0.88) and deployed Word2Vec inference through a Flask REST API integrated with a React frontend.

Word2Vec NLP Recommendation Systems A B Testing Cosine Similarity Flask API React