Prompt Engineer / AI Data & QA Specialist
Zensar Technologies Pvt. Ltd, Pune · Oct 2024 – Present
- Contributed to more than 9+ diverse projects supporting NVIDIA's ongoing and upcoming AI initiatives and data factory operations.
- Worked on NVIDIA AV Alpamayo dataset creation, NVIDIA Cosmos for autonomous vehicle simulation training, and NVIDIA GTC AI model data simulation.
- Generated structured question-answer datasets from visual and textual STEM inputs, supporting multimodal AI training pipelines.
- Performed dense video captioning and temporal event annotation, identifying key actions, timestamps, and contextual scene descriptions.
- Executed document intelligence tasks by extracting educational STEM questions from textbook datasets.
- Designed multi-category questions from physics-driven and egocentric video datasets, supporting physical AI workflows.
- Validated speech datasets by aligning audio recordings with transcripts to improve ASR model accuracy.
- Created human-written captions and transcripts for speech recognition training datasets.
- Annotated multimodal video datasets using bounding boxes, object tracking, and event tagging in SuperAnnotate.
- Served as a Quality Analyst (QA), reviewing annotations, assigning quality ratings, and ensuring adherence to detailed project guidelines.
- Audited annotation outputs from annotators and QC teams to maintain dataset consistency.
- Acted as a Subject Matter Expert (SME), guiding a team of 40 contributors on annotation workflows and technical decisions.
- Evaluated model-generated annotations, improving dataset reliability and model alignment.
- Contributed to initiatives that improved Al model performance by approximately 16%.