Halley Patel

AI Product Manager / Product Engineer with hands-on experience building and evaluating AI systems from problem definition through working prototypes and release validation. Experienced in AI evaluation, benchmarking, prompt engineering, agentic workflows, model monitoring, product requirements, and technical product delivery. Built evaluation frameworks using 712 expert-reviewed cases, defined model quality metrics and regression tests, and evaluated AI systems across 7,500+ images.

Strong technical foundation in Python, SQL, APIs, FastAPI, LLM APIs, RAG, and AI agents, with experience translating ambiguous problems into measurable product requirements and evaluation criteria.

Data Scientist - Radinate - Healthcare AI startup - Ahmedabad, India

(2026-03)

• Owned product definition for an AI decision-support workflow, translating an ambiguous operational problem into structured product requirements, evaluation criteria, and a machine-evaluable workflow.
• Translated domain guidelines and business rules into AI product requirements and system architecture, balancing accuracy, explainability, usability, and implementation constraints.
• Defined product success metrics and evaluation criteria for assessing AI-supported authorization outcomes against standardized evidence.
• Worked directly with domain SMEs and engineering teams to resolve accuracy, usability, and implementation trade-offs and move the workflow toward production delivery.
• Proactively analyzed production-model performance gaps across 15,000+ clinical records, identifying a 53% miss rate on high-risk cases and using structured evaluation to inform a redesigned risk framework.
• Redesigned the model-risk framework, reducing false positives by 52% through revised evaluation criteria and a more systematic approach to performance assessment.
• Established a continuous third-party vendor monitoring workflow by defining confidence-based flagging criteria, supporting repeatable governance and ongoing quality review by clinical risk teams.

Data Science & ML Intern - Radinate - Healthcare AI startup - Ahmedabad, India

(2025-09 - 2026-03)

• Established model-performance success metrics of 0.90 / 0.87 AUROC and tracked post-deployment outcomes, translating technical confidence scores into decision-ready monitoring signals for clinical stakeholders.
• Evaluated a vision-language model across 7,500+ images, identifying output-reliability gaps and driving requirement improvements that expanded structured extraction from 9 to 16 clinical entities.
• Enabled reliable AI release decisions through a benchmarking framework built on 712 expert-reviewed ground-truth cases, defining quality metrics and regression tests across model versions.
• Built a data pipeline converting unstructured inputs into decision-ready structured datasets with 0.87 F1, enabling downstream product experimentation and evaluation at scale.

Diploma - Data Science and Applications - Indian Institute of Technology Madras (2023 - 2025)

Bachelors of Science - Data Science - Gujarat University (2022 - 2025)