Yeraly Gainulla

First Author and Machine-Learning Researcher

SemEval-2026 Task 13, Subtask C

International AI-generated code detection task where teams evaluate systems for detecting code written by AI.

Designed a long-file code detection system and first-authored the paper for a team that ranked first among 50+ SemEval teams.

• Ranked 1st among 50+ teams, competing against AI labs and PhD students; macro-F1 0.7855, 5.2 percentage points above the runner-up. • Designed a Dynamic Multiple Instance Learning pipeline for long files with structural/stylometric features; first-authored the peer-reviewed paper YoungDSMLKZ at SemEval-2026 Task 13: MIL-UniXcoder with Meta-Stacking and Handcrafted Features for AI-Generated Code Detection, published in the ACL Anthology.

Independent Researcher and Developer

KazBERT

Open research project focused on a Kazakh language model, with released weights, tokenizer, code, data, and a technical report.

Trained and released a Kazakh language model, tokenizer, data, and code, reaching 25,000 downloads and use in NLP research across Kazakhstan.

• Continued masked-language-model training for Kazakh and trained a tokenizer; built the Kaggle GPU workflow from data preparation through evaluation. • Released weights, tokenizer, code, data, and the technical report KazBERT: A Custom BERT Model for the Kazakh Language. Reached 25,000 downloads and 15 likes; used in NLP research across Kazakhstan. Extended to sentence similarity, NER, and benchmarks.

Summer Researcher

Nazarbayev University

A research university in Kazakhstan, with research and teaching across science, engineering, and other academic disciplines.

Reviewed AI-based virtual cell systems and predictive biological modelling across animal, plant, and microbial systems.

• Reviewed AI-based virtual-cell systems and predictive biological modelling across animal, plant, and microbial systems.

Chief Technology Officer

Sentra

Edge-first retail analytics platform for local camera ingestion, computer-vision pipelines, event processing, dashboards, and operational risk signals.

Leads technology for an edge-first retail analytics platform. The startup reports KZT 1.5M in cumulative prize funding.

• Lead an edge-first retail analytics platform: local camera ingestion, computer-vision pipelines, event processing, dashboards, and operational risk signals for privacy-sensitive, low-connectivity settings. • Sentra competition results: 3rd place, KZT 300,000 at Astana Hub Pizza Pitch (12 finalists from 141 applications); 1st place, Regional Startups Competition (Auyl Hackathon), Astana; 2nd place, National Startups Competition (Auyl Hackathon); Best Idea, Alga Pitch Day. Sentra reports KZT 1.5M cumulative prize funding.

Head Coder

FIRST Tech Challenge

A student robotics competition where teams design, build, and program robots to complete season-specific challenges.

Develops autonomous routines, path planning, hardware control, and competition software for a robotics team recognized with the Control Award.

• Develop autonomous routines, path planning, hardware control, and competition software; software and control work recognized with the Control Award at Bilim Shyny FIRST Off-Season 2026.

High School

Aqbobek Lyceum