Eric Syetua

Key Skills

Software

Labelbox

CVAT

Scale AI

Appen

Top Subject Matter

LLM Alignment and Multilingual Reasoning

Computer Vision for Robotic Object Picking

Generative Media (Image/Video) Evaluation

Top Data Types

Text

Image

Video

Audio

Top Task Types

RLHF

Object Detection

Segmentation

Transcription

Freelancer Overview

Multilingual AI Trainer & SME. Brings 3+ years of professional experience across complex professional workflows, research, and quality-focused execution. Core strengths include Labelbox, CVAT, and Stable Diffusion. Education includes Master of Science, Mount Kenya University (2023). AI-training focus includes data types such as Text, Image, and Video and labeling workflows including RLHF, Object Detection, and Evaluation.

Labeling Experience

Generative Image/Video Model Evaluator

Image

Additionally, I benchmarked generative image and video models for fidelity and prompt adherence. My focus was on evaluating pixel-level consistency and naturalness in AI-created media. This work supported model improvements for generative tasks. • Benchmarked text-to-image and text-to-video generative models. • Assessed outputs for prompt adherence and detailed visual quality. • Evaluated consistency in generated images and videos at the pixel level. • Helped refine model architecture for generative AI tasks.

Additionally, I benchmarked generative image and video models for fidelity and prompt adherence. My focus was on evaluating pixel-level consistency and naturalness in AI-created media. This work supported model improvements for generative tasks. • Benchmarked text-to-image and text-to-video generative models. • Assessed outputs for prompt adherence and detailed visual quality. • Evaluated consistency in generated images and videos at the pixel level. • Helped refine model architecture for generative AI tasks.

2023 - Present

Computer Vision Data Annotator

CVATImageObject Detection

In the same role at Outlier AI, I annotated complex visual datasets for robotic object picking. This involved using computer vision tools to define spatial parameters and precise grasp points. The process enhanced robotic perception for logistics automation. • Annotated images for object picking using bounding and 3D spatial geometry. • Labeled grasp points and depth perception features for robotics datasets. • Utilized advanced CV tools to segment and classify objects in images. • Enhanced automation and accuracy in warehouse robot tasks.

In the same role at Outlier AI, I annotated complex visual datasets for robotic object picking. This involved using computer vision tools to define spatial parameters and precise grasp points. The process enhanced robotic perception for logistics automation. • Annotated images for object picking using bounding and 3D spatial geometry. • Labeled grasp points and depth perception features for robotics datasets. • Utilized advanced CV tools to segment and classify objects in images. • Enhanced automation and accuracy in warehouse robot tasks.

2023 - Present

Multilingual AI Trainer & SME

LabelboxTextRLHF

As a Multilingual AI Trainer & SME at Outlier AI, I performed RLHF and supervised fine-tuning (SFT) for LLMs in five languages. My work ensured cultural nuance, grammatical precision, and logical accuracy in model responses. I also evaluated and rated generative models for linguistic and idiomatic correctness. • Conducted cross-lingual model alignment using RLHF and SFT. • Ensured high-quality translations and localized reasoning across multiple languages. • Evaluated model outputs for idiomatic accuracy and cultural context. • Reduced translationese and improved natural language generation.

As a Multilingual AI Trainer & SME at Outlier AI, I performed RLHF and supervised fine-tuning (SFT) for LLMs in five languages. My work ensured cultural nuance, grammatical precision, and logical accuracy in model responses. I also evaluated and rated generative models for linguistic and idiomatic correctness. • Conducted cross-lingual model alignment using RLHF and SFT. • Ensured high-quality translations and localized reasoning across multiple languages. • Evaluated model outputs for idiomatic accuracy and cultural context. • Reduced translationese and improved natural language generation.

2023 - Present

Audio Annotator and Transcriber

AppenAudioTranscription

In the capacity of Linguistic Data Consultant, I managed phonetic and semantic labeling for audio recordings. This included diarization, phonetic annotation, and transcription of multi-speaker files in French and Spanish. The output contributed to training STT models with high dialectal accuracy. • Transcribed audio for phonetic and semantic accuracy. • Annotated speaker turns and dialectal variations. • Labeled multi-speaker, multi-dialect content for STT development. • Enhanced language models for under-resourced languages.

In the capacity of Linguistic Data Consultant, I managed phonetic and semantic labeling for audio recordings. This included diarization, phonetic annotation, and transcription of multi-speaker files in French and Spanish. The output contributed to training STT models with high dialectal accuracy. • Transcribed audio for phonetic and semantic accuracy. • Annotated speaker turns and dialectal variations. • Labeled multi-speaker, multi-dialect content for STT development. • Enhanced language models for under-resourced languages.

2022 - 2023

Linguistic Data Consultant (Multimodal)

Scale AIVideoSegmentation

As a Linguistic Data Consultant, I performed frame-by-frame video and audio segmentation to label training data for autonomous systems. This work involved segmenting multimodal data and annotating both acoustic and visual events. The labels enabled improved perception and reaction for AI-driven systems. • Labeled frames in video and segmented sequences for AV perception. • Annotated audio and synchronized it with corresponding video events. • Supported training for autonomous vehicle and system AI. • Enhanced multi-modal datasets for sensor fusion tasks.

As a Linguistic Data Consultant, I performed frame-by-frame video and audio segmentation to label training data for autonomous systems. This work involved segmenting multimodal data and annotating both acoustic and visual events. The labels enabled improved perception and reaction for AI-driven systems. • Labeled frames in video and segmented sequences for AV perception. • Annotated audio and synchronized it with corresponding video events. • Supported training for autonomous vehicle and system AI. • Enhanced multi-modal datasets for sensor fusion tasks.

2022 - 2023

Education

Work History