Олександр

Oleksandr Smutniak
Kamianets-Podilskyi, Ukraine, [відкрити контакти](див. вище в блоці «контактна інформація»)

E-mail: [відкрити контакти](див. вище в блоці «контактна інформація»)

LinkedIn: LinkedIn

Data Labeling Specialist / AI Content Quality / Data Curation

SUMMARY: Detail-oriented AI Data Specialist with 3 years of experience in
high-complexity AI training and Multimodal data annotation. Expert in 3D
Body Node Segmentation (BNS), Human Pose Estimation, and detailed text-
based action description for Tier-1 tech companies (e.g., Snap). Strong
background in RLHF, text-to-video evaluation, and handling complex edge
cases.

TECHNICAL SKILLS:
Data Annotation: Bounding Boxes, Keypoints Annotation (Facial
Landmarks), Response Rewriting, Semantic and Polygonal Segmentation,
Supervised Fine-Tuning (SFT) Data Prep, Object Detection, Text-to-Video
verification, Model Ranking, AI Alignment, Prompt Evaluation, Data Tagging,
Action Description, Human Behavior Labeling, RLHF, AI Hallucination
Identification & Correction, Ground Truth Creation, Vision-Language Model
(VLM) Alignment.

Tools: Google Workspace, Labelbox, Data Annotation Platforms, Snap
LBLR, Label Studio.
Core Competencies: High typing speed, Technical documentation analysis,
Image and video description, Attention to detail, Complex edge case analysis,
Spatial reasoning, English-language content generatioN

PROFESSIONAL EXPERIENCE:
Project Data Annotator | Label Your Data | April 2026 – Present

Performed high-precision data annotation and semantic labeling within
3D environments using Label Studio.
Specialized in complex action recognition tasks, providing detailed
English-language descriptions of human movements and actor
behaviors to train computer vision models.
Executed polygonal segmentation and object tracking with a focus on
maintaining high quality and strict metadata consistency.
Analyzed complex edge cases in human behavior, writing highly
detailed, structured English prompts and descriptions to improve the
model's spatial reasoning and action recognition capabilities.

Data Labeling Specialist | ML Platform Department | Snap (Remote)
January 2025 – March 2026
Annotated complex video datasets for AI model training using Bounding
Boxes, Polygonal, Semantic Segmentation, and Keypoints (Facial
Landmarks).
Evaluated and ranked text-to-video generation outputs based on
prompt alignment, visual fidelity, and logic. Applied RLHF
methodologies and provided corrective feedback to refine generative
model performance.
Identified and corrected AI hallucinations in Vision-Language Models
(VLMs) by rewriting inaccurate generated text to ensure perfect
alignment with complex video content.
Authored high-quality Ground Truth data and performed Response
Rewriting for Supervised Fine-Tuning (SFT), directly improving the
model's accuracy, spatial reasoning, and contextual understanding.

Content Annotator | Dean LLC (Remote) October 2023 – January 2025
Created detailed English-language descriptions for image and video
assets.
Ensured consistency and linguistic accuracy in metadata for
international clients.

EDUCATION:
Kamianets-Podilskyi Ivan Ohienko National University
Bachelor’s Degree in Secondary Education (English Language & Literature)
Graduation: 2025

LANGUAGES
English: Upper-Intermediate (B2).
Ukrainian: Native.