Data Annotator

Senior Data Annotator — Pixelwise Systems

San Jose, CA | January 2024 – Present

• Orchestrated the labeling of 850,000 computer vision frames using CVAT, maintaining a 99.8% bounding box accuracy rate for autonomous driving model training.
• Designed a semantic segmentation ontology for a team of 15 labelers, reducing annotation guideline clarification requests by 42% using Labelbox.
• Automated pre-labeling scripts using Python and OpenCV to preprocess raw image datasets, boosting the overall annotation throughput by 35% across 8 major projects.
• Audited 120,000 NLP text snippets annotated by vendor teams, correcting schema deviations to achieve a consensus score of 98% via Cohen's Kappa metric.

Data Annotator — Semantic Data Labs

San Francisco, CA | August 2021 – December 2023

• Labeled 450,000 textual dataset entries for sentiment analysis and entity recognition models, consistently exceeding daily volume quotas by 18%.
• Identified 4,200 edge-case training samples using Prodigy, collaborating with ML teams to resolve class imbalance and improve model F1-scores by 8%.
• Created 5 training modules on active learning practices for onboarding new hires, accelerating the team's average time-to-productivity by 12 days.
• Reconciled labeling discrepancies in 80,000 audio annotation files utilizing Audacity, improving audio model phoneme recognition accuracy by 14%.

Image annotation, Video annotation, Text labeling, Audio transcription, Semantic segmentation, Bounding box, Polygon labeling, Named Entity Recognition (NER), Sentiment analysis, Data quality assurance, Active learning, Labelbox, CVAT, Prodigy, Roboflow, Label Studio, Python, SQL, Git

Multimodal Autonomous Driving Dataset

Role: Lead Annotator

Tools: CVAT, Python, Roboflow

Managed the semantic segmentation of 250,000 video frames, improving model object-detection accuracy by 12% and finishing the project 18 days early.

NLP Sentiment Fine-Tuning

Role: Data Annotator

Tools: Prodigy, Python, Jupyter Notebooks

Labeled 50,000 domain-specific tweets for fine-tuning LLMs, which raised training data accuracy by 9% and enabled model deployment 2 weeks ahead of schedule.

• Certified Data Annotator (CDA) (2022)
• AWS Certified Cloud Practitioner (2024)

• Languages: English (Native), Spanish (Professional working proficiency)
• Volunteer Work: Open-source data labeler for environmental conservation AI models (2023-present)
• Availability: 2 weeks notice