Yaroslav

Yaroslav Pristupa
Kyiv, Ukraine | [відкрити контакти](див. вище в блоці «контактна інформація») | Email | GitHub

AI & Prompt Engineering Specialist
My passion is in engineering the AI tools that empower creative teams, a skill I honed while pioneering generative AI integration as
a team lead. I moved beyond simple application to focus on what makes models work – and what makes them fail.

My experience is practical: I've used everything from LLMs to Stable Diffusion for text, audio, and image tasks. I specialize in
systematic prompting (XML, Chain-of-Thought) and parameter tuning to get predictable, high-quality outputs. Now, I'm looking to
shift full-time into an R&D environment where I can use my skills in scripting and failure analysis to build truly intelligent and
reliable AI solutions.

Core Competencies
AI / ML:
LLM & Prompt Engineering:
Models: Gemini, ChatGPT, Claude, Llama, Qwen, Grok.

Techniques: System Prompts, Few-Shot Prompting, State Management, Chain-of-Thought.
Skills: Structured Prompting (XML, MD), Performance Optimization (Parameter Tuning: Temperature, Top P,
Thinking Budget), Failure Case Analysis, API Integration.

Multimedia Content Generation:
Image: Stable Diffusion (LoRA), Dataset Preparation & Curation.
Audio: Text-to-Speech (TTS), Music Generation.
Video & Text: AI-assisted content creation and editing.
AI Agents & Architectures:
AI Agents: Foundational knowledge of AI agents and their concepts.
Architectures: RAG (Retrieval-Augmented Generation) - theoretical understanding.
Technical Skills:
Scripting: Python (basic: task automation, data processing, office tasks).
Databases: SQL (basic queries).
Tools: Git, GitHub, LMStudio, Jira, Confluence.
Processes & Soft Skills:
Strong Logical Thinking & Problem-Solving

Cross-Functional Communication
Team Leadership & Mentorship
Task Decomposition & Management
Excellent Written Communication & Documentation

Professional Experience
Team Lead, Art Department | SciPlay | 2020-2025

Led a team of 6 specialists, overseeing the full production lifecycle for 2D/3D assets.
Pioneered and executed the integration of AI tools into the department's workflow. Independently researched and tested
various AI models (LLM, Stable Diffusion), trained the team on their application, and personally handled the fine-tuning of
models (LoRA) for specific tasks.
Decomposed high-level creative goals into actionable technical tasks, implementing an iterative feedback system to ensure
results met stringent quality, style, and tone requirements.

Facilitated effective cross-functional communication between the art department, game designers, and developers.

Projects
Fine-tuning Stable Diffusion for Custom Image Generation

Managed the end-to-end dataset preparation pipeline, including collection, filtering, cropping, and detailed captioning of
hundreds of images.
Successfully fine-tuned models using the LoRA technique to generate images in a highly specific, custom art style.

Automated Metadata Extraction Script | Python

Developed and debugged a Python script to automate metadata extraction from large batches of images, structuring the data
for further analysis.
*Projects and work samples are available on my GitHub.

Languages
English: Fluent (C1)
Ukrainian: Native
Russian: Native