Skip to main content
Carol Nganga

Carol Nganga

AI Trainer | Multimodal Data & Annotation Specialist

Talent
NairobiMember since June 19, 2026
Segmentation maskingFinance-domain prompt engineering
·

Hire this person

Send a job offer directly to this candidate

Segmentation maskingFinance-domain prompt engineering

About

AI Training and Data Annotation Specialist with experience in multimodal evaluation, computer vision annotation, speech synthesis assessment, and RLHF. Skilled in bounding box annotation, segmentation masking, image preference evaluation, finance-domain prompt engineering, and instruction-following model tuning. Detail-oriented and adaptable, with a strong track record of producing high-quality annotations and evaluations across diverse AI training pipelines.

Experience

Image Annotation Specialist at iMerit (2026-03 – Present)

  • Evaluate pairs of AI-generated images against text prompts, selecting the superior output based on prompt adherence, anatomical accuracy, and aesthetics.
  • Assess both explicit and implicit prompt intent, flagging visual errors such as anatomy issues, lighting inconsistencies, and composition problems.
  • Maintain high accuracy under timed conditions, contributing to model quality improvement through consistent, well-reasoned preference judgments.

Data and Image Annotation Specialist at Abaka AI (2025-11 – Present)

  • Trained and evaluated generative AI systems by assessing image outputs against detailed text prompts.
  • Verified prompt adherence, visual accuracy, and consistency across AI-generated images.
  • Conducted comparative analysis and ranked images based on quality, relevance, and prompt alignment.
  • Applied quality control standards to improve model performance and reduce annotation errors.

AI Trainer & Team Lead at Turing (2025-03 – 2025-12)

  • Led SFT and RLHF initiatives for financial-domain language models, enhancing instruction-following, multi-step reasoning, and quantitative analysis.
  • Conducted rigorous evaluation, red-teaming, and error analysis to reduce hallucinations and improve model robustness.
  • Designed annotation frameworks, scoring rubrics, and prompt engineering strategies to optimize alignment and dataset integrity.
  • Mentored AI trainers and enforced data governance and privacy standards.

AI Trainer at RWS (2026-01 – 2026-04)

  • Evaluate voice cloning outputs for speaker likeness, pronunciation accuracy, prosodic consistency, and overall audio naturalness.
  • Perform quality analysis on synthesized speech, identifying phonetic errors, articulation inconsistencies, background noise, and alignment issues.
  • Draw bounding boxes on visual datasets to support object detection and classification model training.
  • Apply segmentation masks to furniture and clothing images, performing pattern removal and surface editing to support generative model training.
  • Provide structured feedback within supervised training pipelines to drive iterative model improvements.

Executive Assistant & Bookkeeper at Freelancer (2024-06 – 2025-02)

  • Managed complex travel arrangements, calendars, and executive inboxes.
  • Handled financial administration including budgets, reimbursements, and expense reports.
  • Coordinated projects and prepared reports and presentations for leadership.

Education

Bachelor of Education Arts in Mathematics & Business Studies – Kenyatta University (2018 – 2022)

Certificate in Virtual Assistance – ALX Foundation (2024-09 – 2024-11)

Skills

AI TrainerMultimodal Data & Annotation Specialist

Reviews