Experience
Experience & Resume
AI and Data professional based in Munich. MSc Informatics from the Technical University of Munich, with experience spanning applied LLM-reliability research at Fraunhofer IKS to scalable ETL pipelines and RAG applications in industry. IEEE-published research in computer vision and NLP, plus awards from ESG-focused AI hackathons.
Career
Work experience.
May 2025 - Feb 2026
Master's Thesis AI Researcher
Fraunhofer IKS
Applied research on the reliability of Large Language Models in safety-critical requirements engineering at Fraunhofer IKS, Munich - from study design and ethics approval through to data analysis.
- Designed and executed end-to-end AI experimentation evaluating LLMs in safety-critical requirements engineering, from study design and ethics approval to data analysis.
- Built and deployed a custom experimental platform using Streamlit to compare LLM-assisted vs. human-only workflows under controlled conditions.
- Analyzed LLM impact on correctness, efficiency, user trust, and learning; identified performance gains alongside risks of over-reliance, dependent on user expertise.
Oct 2024 - Mar 2025
certum.ai
Data / ML Engineer (Working Student)
Certum.ai
Built and scaled data pipelines and NLP retrieval systems to turn unstructured ESG and sustainability reports into structured, queryable data at Certum.ai, Munich.
- Developed and optimized scalable ETL pipelines and API integrations to process ESG and sustainability data from German stock-exchange-listed companies.
- Designed and automated a web-scraping pipeline to extract corporate sustainability reports, achieving 95% data-extraction accuracy.
- Implemented data-orchestration workflows for end-to-end ingestion, processing, and storage of large-scale ESG datasets.
- Built information-retrieval and search-algorithm pipelines to identify emissions and sustainability metrics from unstructured corporate reports.
Dec 2022 - May 2024
AI Engineer (Working Student)
Retorio
Fine-tuned, optimized, and evaluated multilingual language models for a behavioral AI platform at Retorio, Munich - expanding language coverage and production reliability.
- Fine-tuned and deployed multilingual language models supporting 4 new languages, improving inference accuracy, robustness, and global platform coverage.
- Optimized model architectures and training pipelines, reducing inference latency and improving per-task performance by 5-10% in production.
- Built scalable Python evaluation pipelines to benchmark candidate models, track metrics, and auto-detect performance regressions in production.
- Improved system efficiency by 10-15% through targeted ML optimization, automated testing, and performance monitoring.
Jun 2021 - Dec 2022
Data Scientist
Miuul
Delivered end-to-end AI/ML projects for enterprise clients including Siemens, Turkish Airlines, and Skoda - from modeling through to production.
- Led and executed end-to-end AI/ML projects for enterprise clients (Siemens, Turkish Airlines, Skoda), delivering scalable ML solutions to optimize operations and enable data-driven decisions.
- Designed, trained, and deployed production-ready ML and deep-learning models, contributing to model productionization, scalability, and performance optimization.
- Developed and evaluated models across time-series forecasting, NLP, CRM analytics, and predictive modeling, choosing architectures aligned with business/engineering constraints.
Sep 2020 - Sep 2022
Data Science Teaching Assistant
Data Science School (Veri Bilimi Okulu)
Mentored aspiring data scientists through hands-on Python, machine learning, and end-to-end data projects at Data Science School, Istanbul.
- Mentored 200+ participants in Python, data science, and machine learning, supporting end-to-end data projects and production-oriented workflows.
- Applied and explained AI/ML models (data preprocessing, feature engineering, model evaluation) through real-world use cases.
Oct 2020 - Mar 2022
Undergraduate Research Assistant
Istanbul Bilgi University
Built one of the largest Turkish-language image-captioning datasets and benchmarked deep-learning models for multimodal (computer vision + NLP) research at Istanbul Bilgi University.
- Developed a Turkish image-captioning system (computer vision + NLP), delivering one of the largest Turkish-language multimodal benchmarks.
- Extended MS COCO 2014 with 616,767 human-verified Turkish captions across 123,287 images.
- Trained and evaluated 5 deep-learning models (CNN+LSTM ResNet101/152, Meshed-Memory Transformers) for Turkish outputs.
- Led a team of 20+ interns, overseeing data quality, task distribution, and model readiness.
Background
Education.
2025
MSc Informatics
Technical University of Munich (TUM)
10/2022 - 12/2025. Thesis: "Reliability of Large Language Models in Safety-Critical Processes." Final grade 2.4 (German system, 1.0 best).
2022
BSc Computer Engineering
Istanbul Bilgi University
09/2018 - 07/2022. Thesis: "Image Captioning in Turkish Language." Final grade 3.3/4.0.
2019
Business English & C1 Course
University of Liverpool
07/2019 - 08/2019. Business English & C1-level Activate English Language Course.
Recognition
Honours & awards.
2022
HackBoğaziçi Hackathon - 2nd Prize
Boğaziçi University IT Club
Built an automated ESG scoring system analyzing text and image data scraped from four sources; placed 2nd among 14 teams.
2022
Honor Certificate
Istanbul Bilgi University
Awarded for academic excellence and active participation; graduated with distinction.
2021
"AI in Investing" Hackathon - 2nd Prize
CFA Poland
Built an ESG solution measuring gender diversity and analyzing news sentiment (web scraping, computer vision, NLP); ranked 2nd among 44 teams from 28 universities.
Research
Publications.
2022
A complete human-verified Turkish caption dataset for MS COCO and performance evaluation with well-known image caption models
ICECCME 2022 (IEEE)
Expanded the MS COCO dataset with Turkish-language captions, achieving state-of-the-art results in Turkish image captioning using Computer Vision and NLP.
Ready to work together?
Let's build something great. Reach out or take a look at recent projects.