Mustafa Durmuş

Data Scientist dedicated to developing innovative AI solutions.

Istanbul, Turkey

LinkedIn
MD

About

I'm a Senior Data Scientist with 6+ years of experience building production NLP and LLM systems. At Nesine, I designed and built a multi-model customer support chatbot from scratch, combining LLMs, BERT classifiers, and rule-based flows. Before that, I led the AI team at Albert Health, where we built a conversational AI platform for healthcare and published research on how people use chatbots in digital health. I focus on systems that work reliably in production: fast, measurable, and cost-efficient.

Skills

ML
NLP
Data Science
Conversational AI
LLMs
RAG
Docker
Cloud Platforms

Work Experience

Nesine
Hybrid

Sep 2024 - Present

Senior Data Scientist

Designed and built Nesine's customer support chatbot end to end: a multi-model architecture combining a Qwen-based intent classifier, Rasa rule-based flows, a BERT classifier, and LLM-powered help bot. Also responsible for building, validating, and maintaining ML models on large datasets with data and backend engineers.

Albert Health
Hybrid

Jun 2024 - Sep 2024

Senior AI & Data Analytics Engineer

I led the AI team at Albert Health, where we developed a conversational AI platform for healthcare. Additionally, I worked on Large Language Models (LLMs), Retrieval-Augmented Generation (RAG), prompt engineering, and optimizing data pipelines.

Albert Health
Hybrid

Dec 2020 - Jun 2024

NLP Engineer

My responsibilities included overseeing the development of AI and NLP pipelines, including intent classification and entity extraction. I also managed the CI/CD pipeline for our AI models.

Hummingdrone
Hybrid

Jun 2019 - Dec 2020

Robotics Developer

I was a member of the Robotics Development Team, where we focused on creating data-driven drones using approaches such as Machine Learning, Computer Vision, Visual SLAM, Amazon Web Services, and open-source robotics technologies.

Education

Galatasaray University

2023 - 2025
Master Degree in Data Science. Courses: Probability and Statistics, Data Lakes and Data Engineering, Machine Learning and LLMs

Karabuk University

2020 - 2023
Master Degree in CS. Courses: Machine Learning, NLP, Graph Theory, Data Mining

Karabuk University

2014 - 2019
Bachelor's Degree in Computer Science. Thesis:A Neural Image Caption Generator

Projects

Vector Databases and RAG

This project aims to store the vectors of the documents in the database and use them in RAG model.

Mongo DB
Langchain
OpenAI

Principal Component Analysis

Introduction to PCA and its implementation in Python.

Python
Sklearn
Jupyter Notebook

Kolmogorov-Arnold Networks

Explanation of Kolmogorov-Arnold networks.

Medium

Research

User Engagement with A Multimodal Conversational Agent for Self-Care and Chronic Disease Management: A Retrospective Analysis2025
Retrospective analysis of ~24,500 users examining voice vs. screen interaction and engagement predictors.
The Use of Conversational Agents in Self-Management: A Retrospective Analysis2024
A retrospective look at conversational agent usage patterns in a mobile health setting.
Exploring vocal biomarkers as non-invasive fine-tuning assays of cardiovascular health: heart failure model2024
Investigation of vocal features as biomarkers for cardiovascular health.
Conversational agent engagement patterns among individuals with MS: A retrospective analysis of the mHealth application2024
Analysis of conversational agent engagement patterns in multiple sclerosis users.
Descriptive analysis of conversational agent usage characteristics in an asthma app2023
Usage characteristics of a conversational agent deployed in an asthma self-management app.
A rule-based named entity recognition of SNOMED Clinical Terms in Turkish clinical reports2022
A rule-based NER approach for mapping Turkish clinical text to SNOMED codes.

Press ⌘J to open the command menu