AI RESEARCHER · NOTTINGHAM → EDINBURGH

Ali Mert Karaoglu

 

First-Class Honours graduate in Computer Science with AI from the University of Nottingham, building reliable, transparent, and human-centered AI systems — at the intersection of LLMs, NLP, Explainable AI, and AI Safety. This September, starting an MSc in Cognitive Science at the University of Edinburgh.

SCROLL ↓

Who I am

I'm interested in AI research and engineering, with a particular focus on building reliable, transparent, and human-centered artificial intelligence systems. My interests lie at the intersection of Large Language Models (LLMs), Natural Language Processing, Explainable AI, AI Safety, and trustworthy machine learning.

I recently graduated with First-Class Honours in Computer Science with Artificial Intelligence from the University of Nottingham. During my undergraduate studies, I conducted research on zero-shot learning for financial market prediction using LLMs and financial news. Alongside research, I enjoy building practical AI systems that bridge academic ideas with real-world applications — from financial risk prediction, to multimodal human-robot interaction using ROS 2 and LLMs, to explainable AI pipelines for decision support.

My long-term goal is to contribute to frontier AI systems that are not only capable, but interpretable, reliable, and aligned with human values.

Originally from Istanbul, Turkey, currently based in Edinburgh, United Kingdom, where I'll continue my studies with an MSc in Cognitive Science at the University of Edinburgh this September.

View CV (PDF)

Research

Zero-Shot Learning & Explainability for Stock Market Movement Prediction Dissertation
Final-year dissertation · University of Nottingham
Investigated the effectiveness of zero-shot NLP for short-term stock market prediction using financial news, with a focus on explainability and trustworthy AI. Supervised by Dr Shreyank Gowda.
Zero-shot LearningLLMsExplainable AI
arXiv preprintUnder review — SN Computer Science (Springer Nature)
Read on arXiv →

Selected work

Agent-Based Fire Evacuation Simulation
Mar 2026 – May 2026

Agent-based simulation comparing evacuation guide policies across a 14,400-run factorial experiment.

Python agent-based simulation comparing Beacon, Sweeper, and congestion-aware Router guide policies across a 14,400-run factorial experiment. Applied ANOVA and Tukey HSD to evaluate how intelligent routing affects survival rates across building topologies.
PythonAgent-based ModellingStatistics
Breast Cancer Treatment Outcome Prediction Group
Oct 2025 – Dec 2025

ML pipeline predicting treatment outcomes from clinical variables and MRI radiomics data.

End-to-end ML pipeline using scikit-learn and PyTorch predicting pathological complete response (PCR) and relapse-free survival (RFS) from clinical variables and MRI radiomics. Led classification model development and contributed to regression approaches.
scikit-learnPyTorchHealthcare AI
NLP-Based Chatbot
Oct 2025 – Dec 2025

Intent-classification chatbot handling multi-step Q&A and simulated purchase flows.

Chatbot classifying user intent and handling multi-step interactions — Q&A, game lookup, and a simulated purchase flow — using TF-IDF and cosine similarity for intent matching within a modular pipeline.
NLPTF-IDFPython
My Robot Avatar Group
Sep 2024 – May 2025

ROS 2 multimodal human-robot interface combining voice, gesture, and LLM-driven control.

ROS 2–based multimodal human-robot interface in CoppeliaSim with voice and hand-gesture control. Combined a local Ollama LLM with a rule-based expert system to translate natural language into real-time JSON actions, plus a cross-platform TCP bridge and ARKit AR visualisation on iPad.
ROS 2LLMsCoppeliaSimARKit
Learning-Based Hyper-Heuristic for the Sight-Seeing Problem
Mar 2025 – Apr 2025

Learning-based hyper-heuristic adapting strategy selection for the Sight-Seeing Problem.

HyFlex-compatible domain with instance parsing, solution representation, and low-level heuristics. Designed a learning-based selection hyper-heuristic that adapts strategy during search, tuned for competitive results within a 5-minute runtime budget.
HyFlexHeuristicsOptimisation

Where I've been

Jul 2025 – Aug 2025
Artificial Intelligence and Data Science Assistant Expert (Internship)
BNP Paribas, Istanbul, Turkey
  • Constructed a 12-month credit default prediction model using Python and scikit-learn to enhance lending decision accuracy.
  • Gained hands-on experience applying ML techniques to real-world financial data analysis and model performance assessment.
Sep 2024 – Jun 2025
Team Admin — My Robot Avatar Group Project
University of Nottingham
  • Facilitated communication and collaboration among team members, organised meetings, and managed schedules and deadlines.
  • Ensured project goals were met efficiently while fostering a positive team environment.

Academic background

Education

Sep 2026 – Onwards
MSc Cognitive Science
University of Edinburgh · Incoming
Sep 2023 – Jul 2026
BSc Computer Science with Artificial Intelligence
University of Nottingham · First-Class Honours

Certifications

MedTech AI Certificate Univ. of Nottingham MedTech Society · Mar 2025
Introduction to Generative AI AWS · Jul 2025
Python Programming Essentials Rice University · Jan 2023
Lambda Expressions with Java Coursera Project Network · Oct 2022
Python Data Structures University of Michigan · May 2021
Retrieving, Processing, and Visualising Data with Python University of Michigan · Oct 2021
Data Collection and Processing with Python University of Michigan · Feb 2021
Python Functions, Files, and Dictionaries University of Michigan · Feb 2021
Python Basics University of Michigan · Feb 2021

What I work with

AI / ML

Machine Learning
Deep Learning
Natural Language Processing
Large Language Models
Zero-shot Learning
Explainable AI (XAI)
Model Evaluation

Frameworks / Libraries

PyTorch
Transformers
scikit-learn
Pandas
NumPy
SHAP

Research / Engineering

Experimental Design
Statistical Analysis
Data Analysis
Git
Linux
REST APIs

Languages

Python
SQL
C
Java
Lean 3/4

Other

Haskell
Assembler
HTML / CSS
JavaScript

Explore via terminal

ali@portfolio: ~
Welcome to ali.karaoglu — type help to see available commands.
ali@portfolio:~$
Try: whoami · about · skills · projects · publications · play · blog · contact · help

Get in touch

Feel free to reach out — happy to chat about AI research, collaborations, or anything else.