Skip to content
AM

Adil Mereke

@edonelle

Data scientist

Masters student at Nazarbayev University studying Data Science and working as a research assistant at the Institute of Smart Systems and Artificial Intelligence

Career

Experience: 1 years

Experience

Research Assistant

Institute of Smart Systems and Artificial Intelligence (ISSAI)

An artificial intelligence and smart systems research institute at Nazarbayev University.

Dec 2025 — Present

Built Kazakh-language agentic benchmark data and supported research on edge/cloud LLM routing and model selection.

• Built a two-phase pipeline using the Gemini 3 Flash Batch API for LLM translation and rule-based entity mapping to translate a multi-domain agentic benchmark into Kazakh across Telecom, Task Management, Retail, Entertainment, and Airline. • Created the Entertainment domain from scratch, writing 32 tasks and a Kazakh-language content library covering films, theatre, concerts, venues, and cities. • Built shared glossaries to resolve terminology inconsistencies across task instructions, action arguments, and database records covering 500+ users, 1,000+ orders, and 2,000+ reservations.

Research Assistant – Solving PDEs Using Quantum Computers

Nazarbayev University

A research university in Astana, Kazakhstan.

Nov 2023 — Sep 2024

Computed and analyzed spectral gaps for QUBO matrices, comparing problem formulations using a D-Wave quantum annealer.

• Implemented code to compute spectral gaps for various QUBO matrices using the Shabyt HPC cluster. • Analyzed the impact of different truncation methods on spectral gaps. • Used the D-Wave quantum annealer to compare energy states across problem formulations.

Education

BSc of Mathematics

Nazarbayev University

2022 — 2026

Selected projects

EEG-Based Image Classification

Built a signal-processing and deep-learning pipeline with a three-person team to classify animal versus object visual stimuli from 64-channel EEG recordings using the THINGS-EEG2 dataset and RSVP paradigm. Preprocessed data with 0.1–40 Hz bandpass filtering, −0.2s to +0.8s windowing around image onset, downsampling, occipital/parietal channel selection (17/23 channels), and PCA. Benchmarked GRU...

Skills

PyTorchHugging FaceTransformersNumPypandasscikit-learnMatplotlib