Daniel Gikamati

Key Skills

Software

Argilla

Label Studio

Prodigy

Other

Labelbox

Top Subject Matter

No subject matter listed

Top Data Types

Image

Text

Video

Top Task Types

Evaluation Rating

Prompt Response Writing SFT

Question Answering

RLHF

Text Generation

Freelancer Overview

Electrical Engineer specializing in the creation of high quality AI training data for technical domains. My expertise lies in developing challenging prompts and precise evaluation rubrics to ensure technically accurate, logically sound, and physically valid AI outputs. I have a proven track record of reducing model hallucinations in engineering topics through meticulous attention to unit consistency, assumption validation, and comprehensive error taxonomy. Proficient in the full SFT and RLHF workflow including prompt response writing, pairwise comparison, and gold set validation, I deliver reliable, high impact training data for power systems, electronics, and control theory. My engineering background enables me to not only annotate data but to teach AI the fundamental principles of electrical engineering, ensuring outputs are safe, accurate, and actionable for real world applications.

Labeling Experience

Engineering Q&A Dataset for AI Fine-Tuning

LabelboxTextQuestion AnsweringRLHF

Spearheaded the creation of a high-quality dataset to fine-tune an LLM for technical accuracy in electrical engineering. Authored over 500 challenging prompts and detailed, step-by-step solutions covering power systems, electronics, and circuit analysis. Performed rigorous evaluation and rating of AI-generated responses against a strict rubric focused on unit consistency, assumption validation, and factual correctness. Identified and tagged error types to create a taxonomy for model improvement. The project resulted in a 40% reduction in model hallucinations for engineering topics.

Spearheaded the creation of a high-quality dataset to fine-tune an LLM for technical accuracy in electrical engineering. Authored over 500 challenging prompts and detailed, step-by-step solutions covering power systems, electronics, and circuit analysis. Performed rigorous evaluation and rating of AI-generated responses against a strict rubric focused on unit consistency, assumption validation, and factual correctness. Identified and tagged error types to create a taxonomy for model improvement. The project resulted in a 40% reduction in model hallucinations for engineering topics.

2024 - 2025

Text Summarization and Finalization – Mindrift

MindriftTextText SummarizationDiagnosis

Refined and summarized AI-generated content for clarity, coherence, and tone consistency. Tasks included polishing long-form outputs, simplifying technical language, aligning content with prompts, and ensuring logical flow. Actively contributed to improving output quality of LLMs by flagging redundancies, optimizing readability, and finalizing pieces for client-facing use. Maintained strong internal QA scores and was consistently assigned complex editing sets.

Refined and summarized AI-generated content for clarity, coherence, and tone consistency. Tasks included polishing long-form outputs, simplifying technical language, aligning content with prompts, and ensuring logical flow. Actively contributed to improving output quality of LLMs by flagging redundancies, optimizing readability, and finalizing pieces for client-facing use. Maintained strong internal QA scores and was consistently assigned complex editing sets.

2023 - 2024

Multimodal Annotation – Aligner

OtherImageBounding BoxPolygon

Labeled more than 35,000 data points across image-caption pairs, video sentiment scoring, and multilingual audio transcription. Delivered precise annotations using CVAT and Label Studio for vision and voice datasets. Maintained an average QA score of 99.5 percent across nine unique projects. Contributed tips and internal guides to help new annotators ramp up and stay consistent with quality expectations.

Labeled more than 35,000 data points across image-caption pairs, video sentiment scoring, and multilingual audio transcription. Delivered precise annotations using CVAT and Label Studio for vision and voice datasets. Maintained an average QA score of 99.5 percent across nine unique projects. Contributed tips and internal guides to help new annotators ramp up and stay consistent with quality expectations.

2023 - 2024

LLM Prompt Evaluation – Outlier AI

OtherTextText GenerationText Summarization

Evaluated over 20,000 LLM-generated responses for safety, clarity, and alignment with structured rubrics. Tasks involved scoring completions based on coherence, factuality, and tone, with attention to edge case handling and hallucination detection. Maintained over 98 percent reviewer alignment and contributed feedback that improved evaluation rubrics across

Evaluated over 20,000 LLM-generated responses for safety, clarity, and alignment with structured rubrics. Tasks involved scoring completions based on coherence, factuality, and tone, with attention to edge case handling and hallucination detection. Maintained over 98 percent reviewer alignment and contributed feedback that improved evaluation rubrics across

2023 - 2024

Education

Work History