Christian Armando

Key Skills

Software

CVAT

Labelbox

Mindrift

OneForma

Top Subject Matter

No subject matter listed

Top Data Types

Video

Image

Computer Code Programming

Audio

Top Label Types

Bounding Box

Segmentation

Object Detection

Action Recognition

Tracking

Polygon

Classification

Text Summarization

Geocoding

Computer Programming Coding

Emotion Recognition

Audio Recording

Transcription

Freelancer Overview

I am an experienced AI Training Specialist with over 3 years of hands-on expertise in data labeling and annotation for computer vision and speech recognition projects. My background includes delivering high-accuracy video, image, and audio annotations using tools like Labelbox, CVAT, Supervisely, SageMaker Ground Truth, and YOLO pipelines. I have consistently maintained annotation accuracy above 98% while supporting large-scale projects, such as autonomous vehicle video labeling, e-commerce image segmentation, and speech recognition audio annotation. My skills cover object tracking, segmentation, transcription, and quality assurance, and I am adept at collaborating with ML engineers to refine guidelines and improve model performance. I am committed to delivering precise, reliable datasets that drive the success of AI and machine learning initiatives.

Labeling Experience

Improved dataset cleanliness by identifying and flagging corrupted or low-quality images. Assisted in refining product category taxonomy to reduce classification ambiguity. Delivered export-ready datasets formatted for ML ingestion.

MindriftComputer Code ProgrammingTrackingText Summarization

Supported AI training workflows by developing lightweight Python scripts to clean, preprocess, validate, and structure annotated datasets for machine learning pipelines. Responsibilities included: Writing Python scripts to convert annotation formats (JSON, CSV, XML) for YOLO and other model training frameworks Automating dataset validation checks to detect missing labels, inconsistent bounding boxes, and formatting errors Performing text preprocessing (tokenization, normalization, data cleaning) for NLP datasets Creating structured prompt-response datasets for supervised fine-tuning (SFT) Implementing basic SQL queries to retrieve and verify dataset entries Supporting function-calling evaluation tasks by validating structured outputs Maintained clean, well-documented code to ensure reproducibility and scalability across projects. Quality measures included: Manual cross-verification of automated scripts Testing scripts against sample datasets before deployment

Supported AI training workflows by developing lightweight Python scripts to clean, preprocess, validate, and structure annotated datasets for machine learning pipelines. Responsibilities included: Writing Python scripts to convert annotation formats (JSON, CSV, XML) for YOLO and other model training frameworks Automating dataset validation checks to detect missing labels, inconsistent bounding boxes, and formatting errors Performing text preprocessing (tokenization, normalization, data cleaning) for NLP datasets Creating structured prompt-response datasets for supervised fine-tuning (SFT) Implementing basic SQL queries to retrieve and verify dataset entries Supporting function-calling evaluation tasks by validating structured outputs Maintained clean, well-documented code to ensure reproducibility and scalability across projects. Quality measures included: Manual cross-verification of automated scripts Testing scripts against sample datasets before deployment

2022

Autonomous Vehicle Video Object Detection & Tracking Annotation

CVATVideoBounding BoxSegmentation

Contributed to a large-scale autonomous vehicle dataset focused on improving object detection and tracking accuracy for real-world driving environments. Annotated over 5,000+ video frames per dataset batch, performing: Frame-by-frame bounding box annotation for vehicles, pedestrians, cyclists, and traffic signs Multi-object tracking with persistent IDs across frames Polygon segmentation for complex object boundaries Action recognition tagging (e.g., turning, crossing, stopping) Occlusion and visibility state labeling Used Labelbox and CVAT to ensure precise annotations aligned with YOLO model training requirements. Maintained strict adherence to detailed annotation guidelines and validation checklists. Quality control measures included: Self-review and secondary peer validation Consistency checks across frames Bounding box accuracy tolerance thresholds Regular calibration sessions to align labeling standards Achieved >98% annotation accuracy and consistently met weekly pro

Contributed to a large-scale autonomous vehicle dataset focused on improving object detection and tracking accuracy for real-world driving environments. Annotated over 5,000+ video frames per dataset batch, performing: Frame-by-frame bounding box annotation for vehicles, pedestrians, cyclists, and traffic signs Multi-object tracking with persistent IDs across frames Polygon segmentation for complex object boundaries Action recognition tagging (e.g., turning, crossing, stopping) Occlusion and visibility state labeling Used Labelbox and CVAT to ensure precise annotations aligned with YOLO model training requirements. Maintained strict adherence to detailed annotation guidelines and validation checklists. Quality control measures included: Self-review and secondary peer validation Consistency checks across frames Bounding box accuracy tolerance thresholds Regular calibration sessions to align labeling standards Achieved >98% annotation accuracy and consistently met weekly pro

2023 - 2024

Additional Information (Optional) Delivered YOLO-ready datasets formatted for training pipelines. Collaborated with ML engineers to refine annotation guidelines based on model error feedback. Experienced working with high-resolution urban driving footage in varied lighting and weather conditions. Comfortable handling large-scale video datasets in fast-paced remote environments.

LabelboxImageBounding BoxPolygon

Worked on a large-scale e-commerce dataset focused on improving product detection, background removal, and recommendation system accuracy. Annotated 10,000+ high-resolution product images, performing: Polygon segmentation to isolate products from complex backgrounds Bounding box labeling for object detection models Multi-object instance segmentation within single images Product category classification tagging Attribute labeling (color, size, orientation) Ensured annotations were optimized for computer vision training pipelines and compatible with YOLO-based detection frameworks. Quality measures included: Multi-pass review process Boundary precision verification for segmentation masks Cross-checking classification labels with taxonomy guidelines Maintaining >99% annotation consistency rate

Worked on a large-scale e-commerce dataset focused on improving product detection, background removal, and recommendation system accuracy. Annotated 10,000+ high-resolution product images, performing: Polygon segmentation to isolate products from complex backgrounds Bounding box labeling for object detection models Multi-object instance segmentation within single images Product category classification tagging Attribute labeling (color, size, orientation) Ensured annotations were optimized for computer vision training pipelines and compatible with YOLO-based detection frameworks. Quality measures included: Multi-pass review process Boundary precision verification for segmentation masks Cross-checking classification labels with taxonomy guidelines Maintaining >99% annotation consistency rate

2023 - 2023

Speech Recognition & Speaker Diarization Audio Annotation

OneformaAudioClassificationEmotion Recognition

Contributed to large-scale speech recognition and conversational AI training projects by annotating and validating diverse audio datasets. Annotated and processed 500+ hours of audio content, including: Clean and verbatim transcription of conversational and multi-speaker recordings Speaker diarization (identifying and labeling individual speakers) Emotion tagging (neutral, happy, frustrated, etc.) Sound event detection (background noise, music, interruptions) Audio segmentation and timestamp alignment Accent and dialect classification Used Audacity for waveform segmentation and noise isolation, and Praat for phonetic and speech pattern analysis. Maintained structured labeling within SageMaker Ground Truth and Label Studio platforms. Quality assurance measures included: Timestamp precision checks Cross-verification of transcription accuracy Adherence to strict transcription guidelines Background noise and non-speech labeling consistency Regular quality audits.

Contributed to large-scale speech recognition and conversational AI training projects by annotating and validating diverse audio datasets. Annotated and processed 500+ hours of audio content, including: Clean and verbatim transcription of conversational and multi-speaker recordings Speaker diarization (identifying and labeling individual speakers) Emotion tagging (neutral, happy, frustrated, etc.) Sound event detection (background noise, music, interruptions) Audio segmentation and timestamp alignment Accent and dialect classification Used Audacity for waveform segmentation and noise isolation, and Praat for phonetic and speech pattern analysis. Maintained structured labeling within SageMaker Ground Truth and Label Studio platforms. Quality assurance measures included: Timestamp precision checks Cross-verification of transcription accuracy Adherence to strict transcription guidelines Background noise and non-speech labeling consistency Regular quality audits.

2022 - 2023

Education

Work History