I'm Arsen Bakhitbekov,
NLP Researcher
Based in Almaty.

Independent AI researcher and developer seeking to align Large Language Models with diverse linguistic frameworks. My primary pursuit is pushing the boundaries of culturally aligned AI—specifically focusing on MENA-region NLP and foundation models like the Jais ecosystem. Actively advancing cross-lingual LLM strategies and building scalable AI products. My ultimate academic goal is to join MBZUAI, contributing to the vanguard of global AI diversity.

🇬🇧🇷🇺🇨🇳🇰🇿🇦🇪

Cross-Lingual LLM Bias Research

Independent Researcher

Sept 2023 - Present

Investigating how LLMs like Jais and Llama change their personality, safety stance, and depth of knowledge when switched from English to Russian, Chinese, or Kazakh. Quantified the 'Safety Curtain' and 'Underrepresented Tax', proving the critical need for region-specific models like Jais to prevent cultural erasure.

Tech: PyTorch • HuggingFace • Jais 30B

View Repository

DragNScroll — EdTech via Short-Form Video

Founder / Technical Lead

Jan 2024 - Present

Building a platform leveraging short videos (TikTok-style) for immersive, context-rich Chinese vocabulary acquisition. Integrating NLP for auto-captioning, difficulty grading, and personalized content recommendation.

Tech: Targeting HSK Learners • AI-driven personalization

Watch Demo

🇨🇳🇬🇧

AI-Driven Business Chinese Platform

NLP Developer

2023

Developed an AI-driven web platform with an integrated NLP chatbot for simulating real-world business negotiations. Vastly improved over traditional static learning by deploying context-rich, simulation-based conversational AI.

Tech: Next.js • Conversational AI • Presented at YDF-2026

View Repository

🇰🇿

PM2.5 Allergy Monitoring ML

Data Scientist

2023

Applied mixed-methods ML to correlate atmospheric particulate matter with public health metrics in Almaty. Modeled environmental IoT data to propose an intelligent, personalized web monitoring platform architecture.

Tech: Python • Scikit-learn • Presented at DKU

View Repository

Platform Engineer & AI Integration Contributor

icomBooster — AI Agent for Sales

Explore

• •Core Platform: Directly contributed to the core product — an AI-powered CRM booster (Next.js + FastAPI + Latenode) that enriches every new CRM signup with 40+ data points including company profile, role, funding stage, and tech stack.
• •AI at Scale: Worked on integrating GPT-4 for automated lead scoring in under 2 seconds — operating at the intersection of real-time data pipelines and production-grade LLM inference.
• •Launch Impact: The platform launched on the Intercom App Store and Product Hunt. Early adopters reported a +40% MRR lift, +18pp increase in qualified leads, and a -25% reduction in CAC — validated B2B outcomes at scale.
• •API Architecture: Gained exposure to 40+ CRM & enrichment API integrations (documented via Swagger), developing a deep understanding of enterprise software ecosystems.

IT Developer & Support (Team Member)

Zerbulak (Largest Waterpark in KZ)

Explore

• •IoT & Infrastructure: Collaborated on integrating smart locker systems and scaling the mobile ticketing ecosystem.
• •System Ops: Contributed to optimizing backend logic to ensure 24/7 high-traffic stability for 1000s of daily visitors.
• •Technical Support: Assisted in maintaining infrastructure reliability and bridging the gap between physical IoT hardware and digital user experience.