Aysel Ozgur

Data-driven AI Evaluation Software Engineer and Technical Lead with 17+ years of experience at Google and Apple designing, validating, and benchmarking complex ranking, natural language processing (NLP), and machine learning systems. Expert at translating high-level organizational policies and safety guidelines into strict automated technical constraints and benchmarking workflows. Proven track record of designing high-throughput side-by-side (SxS) evaluation platforms, defining rigorous grounding and truth metrics, and managing golden benchmarking datasets to safeguard model safety, alignment, and retrieval quality at scale.

Machine Learning & Evaluation Engineer - Apple

(2022-01 - 2023-12)

Siri Web Answers and Ranking

• Diagnosed, refactored, and strengthened core ranking, indexing, and evaluation libraries, optimizing evaluation correctness, benchmark accuracy, and release safety.
• Designed and shipped an automated user-log data processing pipeline in Python and C++ to scale query evaluation datasets, optimizing retrieval efficiency and text-matching metrics.

Software Engineer, Tech Lead - Google

(2019-01 - 2022-12)

Systems Validation & Compliance, Ads Integrity Targeting

• Partnered with cross-functional policy and review teams to translate strict safety regulations into programmatic backend constraints, blocking policy violations across Search and Display channels.
• Built and optimized a TensorFlow-based text and signal classifier that scaled automated data validation and automated review workflows, accelerating ground-truth quality checks.

Software Engineer - Google

(2017-01 - 2019-12)

Quality Infrastructure, Ads Quality, Ads Measurement and Attribution

• Built robust quality infrastructure for geo-ads spend-experiment optimization, running data analysis and data validation workflows across 20 countries to ensure metric alignment.
• Designed and launched a dynamic geo-marketing-areas system and datastore to optimize retrieval accuracy and grounding metrics for Search Ads.

Software Engineer - Google

(2016-01 - 2016-12)

Infrastructure & Data Quality, Display Ads Infrastructure

• Applied predictive machine learning models to classify web content, reducing structural processing noise and establishing automation to safeguard scraping and indexing pipeline efficiency.
• Executed retrieval and text experiments utilizing salient-synonym extraction to systematically benchmark ad targeting relevance.

Software Engineer - Google

(2013-01 - 2015-12)

Search Quality & Content Evaluation, Google Now

• Developed a higher-precision and higher-recall evaluation algorithm to measure the faithfulness and accuracy of user-routine pattern predictions triggering automated contextual actions.
• Built data processing pipelines for personalized handling and semantic evaluation of sensitive localized user query strings.

Software Engineer - Google

(2006-01 - 2012-12)

WebSearch Ranking Evaluation & Localization

• Designed and built a next-generation side-by-side (SxS) ranking evaluation platform that operated 5x faster than its predecessor, optimizing core logic and text evaluation workflows to reduce latency.
• Designed a metro-query synonym framework to optimize local-results ranking accuracy, executing validation cycles across four consecutive production launches.
• Shipped NLP-driven transliteration synonyms for diverse global query streams, ensuring linguistic precision and text matching across complex international traffic.
• Expanded international query language detection capabilities from 30 to over 80 languages, building evaluation templates to verify accuracy across 250 global search interfaces.
• Authored a comprehensive error-diagnostic action template to benchmark global search results quality against external competitors; coordinated 20+ engineers to debug features over an evaluation golden set of 1,000+ random queries.

Ph.D. - Computer Science - University of Minnesota

B.S. - Computer Engineering - Boğaziçi University

Exchange Student - Computer Science - University of Illinois Urbana-Champaign