• Contributed to training AI models in Python, JavaScript, and SQL by reviewing and evaluating coding responses and code outputs by the models.

• Evaluated responses against requirements, ensuring accuracy, completeness, and adherence to instructions.

• Identified and resolved logic errors, edge cases, and inefficiencies through systematic testing and debugging.

• Assigned quality ratings and provided clear, concise justifications based on structured evaluation criteria.

• Conducted comparative analysis and ranking of solutions, documenting results to ensure consistency and transparency.

• Developed complex prompts for AI models in the White Wolf RLHF Project, focusing on creating meaningful deviations between models.

• Improved AI models' ability to accurately follow instructions and respond truthfully in complex scenarios through Reinforcement Learning with Human Feedback, leading to more reliable and contextually appropriate models’ outputs in real-world applications.

Reviewed videos and carefully analysed corresponding audio components.

•

Segmented and created precise visual masks for objects responsible for making sounds within videos.

•

Identified and annotated three visual objects per video that were most likely to produce sound but did not, enhancing model learning accuracy.

•

Ensured high-quality annotations by cross-verifying audio-visual alignment and maintaining adherence to detailed project guidelines.

Annotated short-form and long-form video clips with high-detail scene descriptions, character identification, and precise action sequencing.

Documented physical appearance of subjects (clothing, accessories, hairstyle, body features) and spatial relationships between characters, objects, and camera angles.

Captured motion details including body orientation, facial expressions, and action timing using time-coded annotations.

Followed strict annotation guidelines to avoid subjective interpretation and ensure objective, verifiable descriptions.

• Created instructional captions to guide AI agents in performing website searches and navigation tasks.

• Ensured captions were precise, clear, and aligned with project guidelines for AI training.

• Transcribed and annotated video content to improve AI comprehension of visual and auditory elements.