Predictive ML
Demand forecasting, dynamic-pricing signals, fraud detection, classification and decision support with clear baselines and evaluation.
XGBoostOptunaTime series
Data Scientist · Applied ML & AI
I turn real-world data into decision systems that ship.
I build production-minded systems across demand forecasting, dynamic pricing, recommendations, optimization and applied AI — connecting model quality to operational and financial outcomes.
Retail & marketplace R&D
End-to-end ML delivery
Türkiye
10B+model-training records in the forecasting and dynamic-pricing pipeline
21% lowerRMSE versus the baseline forecasting model
4.5% gainbest objective lift under real warehouse constraints
10+ systemsacross retail, marketplace and operational decision-making
Where I create value
One practice, four connected capabilities.
I combine machine learning, data engineering and operations research so that predictions lead to controlled actions — not isolated notebooks.
Personalization
Candidate generation, collaborative and content-based filtering, sequential patterns and semantic ranking.
RecSysRankingEmbeddings
Decision optimization
Inventory, transfers, batching, slotting, scheduling and simulation under real operational constraints.
ORGenetic algorithmsSimulation
Applied AI systems
LLM, RAG, multimodal and human-in-the-loop workflows designed for evaluation, automation and safe handoff.
RAGAgentic AIMultimodal
R&D delivery portfolio
Applied AI across the customer and operations lifecycle.
A concise view of completed and ongoing systems for D&R and idefix — spanning discovery, risk, visual intelligence, planning and operational decisions.
Recommendations, cross-sell & discovery
Collaborative and content-based filtering, sequential behavior, candidate generation and semantic/LLM-assisted ranking.
Forecasting-driven dynamic pricing
SKU-store demand forecasts feed an optimization layer that recommends prices using competitor, inventory and commercial signals across 20,000+ SKUs.
Agent-based fraud decision support
Fraud prediction, ranked alerts, review queues, audit sampling and feedback datasets for controlled investigation.
Product and catalog intelligence
Product color recognition, sentiment extraction, book metadata enrichment, classification and listing automation.
Product-aligned scene generation
Generative imagery workflows that create brand-appropriate backgrounds aligned with product content and visuals.
Simulation & optimization
Bottleneck analysis, shift scheduling, store simulation, warehouse slotting and order batching under real constraints.
Experience
From data infrastructure to applied AI R&D.
My path combines data engineering foundations with end-to-end ML, optimization and AI systems across retail, marketplace and document intelligence.
Download the full CVMar 2024 — Present
Data Scientist · R&D
Turkuvaz Media Digital · Türkiye
End-to-end projects spanning personalization, demand forecasting and dynamic pricing, fraud detection, computer vision, NLP/LLM, simulation and warehouse optimization.
May 2023 — Mar 2024
Data Engineer
Treo Bilgi Teknolojileri / Treomind · Türkiye
Enterprise HPE Data Fabric and Unified Analytics implementations across distributed data pipelines and ML deployment workflows.
2022 — 2023
AI & Backend Internships
WizardTales · Germany / Texinsight · Türkiye
Deep learning recommender systems for e-commerce and transformer-based NLP/OCR pipelines for document-scanning applications.
Aug 2020 — May 2021
Founder · Commercial Operations
MIRZA TRADE · Türkiye
Built and managed a small textile business serving European markets, covering product development, export operations, KPI tracking, business planning, marketing and brand positioning.
Education
B.A. Business Administration (English)
Istanbul Commerce University
GPA 3.54/4.00, ranked 2nd in department, with academic merit scholarships in 2019–2020 and 2021–2022.
Credentials
Applied AI & Data Science
DeepLearning.AI · IBM · HPE
Deep Learning Specialization, IBM Data Science Professional Certificate and HPE Ezmeral Unified Analytics Software.
Research & writing
Applied research grounded in operational data.
My conference work focuses on scalable forecasting, behavioral sequence mining and warehouse optimization — areas where methodology must survive real constraints.
ISPR 2025 Springer 2026
Retail Sales Forecasting Using Competitor Data
Optuna-XGBoost benchmarked on real retail data with competitor pricing, promotions and temporal features.
Open DOIETMS 2025 Springer
GSP vs PrefixSpan for Sequential Pattern Mining
A production-oriented comparison across runtime, memory, pattern output and multiple retail granularities.
View contributionINFUS & YAEM 2026
Warehouse batching and slotting optimization
Comparative heuristic and metaheuristic studies using real operational order, product and warehouse data.
Explore research