Gokturk Buchholz

About Gokturk Buchholz

I am an Applied Scientist specializing in machine learning and natural language processing, with a focus on deeply understanding data and building models that solve real-world problems. My work blends rigorous analysis, structured experimentation, and practical ML engineering to create reliable, high-impact systems.I use both traditional ML techniques—such as clustering, dimensionality reduction, and classical modeling—and modern NLP approaches, including LLM fine-tuning, embeddings, and retrieval-augmented generation (RAG). This combination helps me uncover structure in complex datasets, engineer meaningful features, and design modeling solutions grounded in evidence.My experience includes developing and evaluating LLM-based workflows, building classification, anonymization, ranking, and decision-support models, and constructing evaluation frameworks for performance, robustness, and model behavior. I take a data-centric approach to modeling—using error analysis, label-quality assessment, imbalance mitigation, and SME collaboration to improve both datasets and model reliability. I ground my work in scientific rigor through careful metric design, Bayesian reasoning, and methodical evaluation.I thrive in collaborative, research-guided environments that support autonomy and thoughtful modeling. I enjoy roles where I can explore data deeply, develop ML-driven solutions, and iterate based on evidence. I am also expanding my expertise in agentic workflows and modern LLM tooling.Skills: Machine Learning, NLP, LLM Fine-Tuning, Embeddings, RAG, Clustering, Dimensionality Reduction, Classical Modeling, Model Evaluation, Feature Engineering, Bayesian Methods, Python, PyTorch, SQL, AWS, Data-Centric Modeling, Error AnalysisEducation: Stanford University — Artificial Intelligence Professional ProgramPublication: Using GPT-3 for Detecting False Political Statements (arXiv)GitHub: https://github.com/marsgokturk

Experience

1. Senior Data Scientist

   Bugcrowd

   Nov 2023 — Present · US

   As a Data Scientist and later Senior Data Scientist, I led applied ML/NLP initiatives focused on understanding real-world data, building modeling solutions, and improving system reliability. My work spanned traditional ML, modern LLM-based approaches, and end-to-end development of production-bound workflows.Key Contributions- Fine-tuned a transformer-based model for a high-cardinality classification task, improving accuracy by ~80% over the few-shot baseline and achieving even higher accuracy on the high-confidence subset (~40% of samples) through calibrated confidence scoring and selective prediction- Developed a few-shot prompting strategy using AWS Bedrock (Batch) to parse highly complex technical jargon and map specific data entities to a structured schema. Achieved 97% model accuracy in entity identification, significantly outperforming traditional NER (Named Entity Recognition) baselines for specialized technical content- Performed deep error analysis on multi-class models, identifying ~10% mislabeled samples and uncovering significant class imbalance; collaborated with SMEs to validate labels, refine evaluation datasets, and apply data-centric remedies (selective sampling, weighted losses) to improve reliability- Designed evaluation frameworks for LLM behavior, including adversarial prompting, faithfulness checks, robustness assessment, and compliance testing- Used traditional ML techniques—clustering, dimensionality reduction, classical modeling—to uncover structure in complex datasets and guide feature engineering- Built predictive and ranking models that improved operational planning and surfaced high-quality signals- Integrated insights from ML/NLP research into modeling strategies, embedding workflows, and evaluation processes to improve system stability and relevance over time.

Education

• University of Washington

  Bachelor of Science (BS), Computer Science and Systems

Skills

• C#
• Xml
• Integration Testing
• Maven
• Spring Boot
• Relational Databases
• Html
• Representational State Transfer (Rest)
• Web Development
• Web Service Development
• D3.js
• Data Modeling
• Web Application Development
• Applications Software Development
• Http
• Software Development Life Cycle
• Ssl Certificates
• Sql
• Node.js
• Java Message Service (Jms)
• Object Oriented Design
• Algorithms
• Https
• Perforce
• Python
• Json
• Junit
• Data Structures
• Javascript
• Amazon Web Services (Aws)
• Git
• Amazon Ec2
• Java
• Unit Testing
• Version Control
• Software Engineering
• Redis
• Computer Science Fundamentals