ORIGINAL BUILD · Learning product
AI Code Arena Quest
The machine-learning practice environment I wished I had: code-first challenges, real datasets, and immediate feedback.
React · Machine Learning · AI Education
KAAN KAYA · DATA SCIENTIST · LUXEMBOURG
I spent a decade moving upstream—from reconciling the number, to building the data platform behind it, to modelling what informs the decision. Today I work across prediction, explainability, and causal inference.
Current focus Platforms · LightGBM · SHAP · DML / DoWhy
PUBLIC WORK
My day-to-day work is private. These projects show how I think about systems, models, and useful products.
ORIGINAL BUILD · Learning product
The machine-learning practice environment I wished I had: code-first challenges, real datasets, and immediate feedback.
React · Machine Learning · AI Education
ORIGINAL BUILD · Data platform
One dataset, end to end—from ingestion and orchestration to modelling and a decision-ready dashboard.
AWS · Airflow · dbt
ORIGINAL BUILD · Data product
A TypeScript workbench for turning tables into graph schemas and repeatable Neo4j loads.
TypeScript · Neo4j · ETL
ORIGINAL BUILD · Applied AI
A local-first retrieval system for grounded questions over my academic notes, using FAISS and private inference.
RAG · FAISS · GPU Acceleration
ORIGINAL EXPERIMENT
COURSE WORK · CLEARLY CREDITED
COURSE WORK · CLEARLY CREDITED
COURSE WORK · CLEARLY CREDITED
WORK HISTORY · 2015—PRESENT
I began in e-commerce data science, then worked through P&L, market risk, regulatory reporting, platforms, and data products. Each step added context I now bring to modelling and causal work.
Eight roles · finance, platforms, and data science
Jan 2025 - Present
Amazon
Building data products and analytical tools for operational decision support.
Data Science · Machine Learning · Explainable AIAug 2024 - Jan 2025
Hubfinance
Led data strategy and implementation for a fintech startup.
NLP · Power BI · PythonMay 2022 - Jul 2024
Schroders
Transformed financial and regulatory reporting processes for a global asset manager.
Financial Reporting · Data Modeling · SDLCAug 2021 - Apr 2022
Lingaro, Carmignac
Delivered data management, automation, and reporting solutions for multiple clients.
Consultancy · Data Management · Regulatory ReportingMay 2020 - Jul 2021
Standard Chartered Bank
Led strategic redesign of market risk reporting systems for a global bank.
Market Risk · Automation · Regulatory ComplianceOct 2018 - Oct 2019
J.P. Morgan
Partnered with trading desks and risk teams on P&L reconciliation and data quality.
P&L · Trading Analytics · Risk MetricsJul 2016 - Sep 2018
NatWest Markets
Contributed to standardization and automation of daily P&L reports.
P&L Reporting · Automation · Data ConsistencyMar 2015 - Jul 2016
FRU.PL
Built early ETL pipelines and dashboards for a high-growth e-commerce business.
ETL · Dashboards · E-commerce AnalyticsHOW I WORK
The model is only useful if the data holds up, the reasoning can be inspected, and the result fits the way people work.
01 / BUILD
At Amazon, I build internal decision-support data products. Business context, scale, and implementation details remain private.
EDUCATION & LEARNING
Finance taught me to respect the number. Computer science is helping me build the systems around it.
2024 - 2027 · IN PROGRESS
2015 - 2017
RECENT CREDENTIALS
WORKING TOOLKIT
Python · SQL · AWS · Airflow · dbt · Spark
GenAI · TensorFlow · PyTorch · scikit-learn · NLP · Pandas
Power BI · Tableau · Looker · QuickSight · Superset · Excel
Based in Luxembourg. I work in English and Turkish, speak Polish conversationally, and am learning Luxembourgish and French.
Away from the screen: electric guitar, piano, and gardening.
CONTACT
If you are working on a messy decision problem—or simply want to compare approaches to explainable ML, causal inference, or data platforms—send me a note.