Hari Krishna Tirunagari

Experience

1. Ai Ml Data Operations Analyst

   Apple

   Nov 2019 — Present · Hyderabad, IN

   LLM - Large Language Models: Fine-Tuning of the LLM’s.Ranking of the responses based on the accuracy, relevance, Concision, truthfulness, Harmfulness of the responses. Describing the reasons for grading. Harmful requests Analysis : Grading of the Harmful requests on how AI has reacted to harmful requests by following the safety guidelines. Successfully met all the quality standards as per the project requirement.Audio Transcription : Transcribing the audio by listening - speech to text.Accuracy : Judging the Siri response and grading them which depends on decision making and judgemental call as per the response.Audio segment annotation: Audio segmentation and Audio Labeling.Face Annotation : Face Annotation and Labeling the video.Helpfulness : Judging the Siri responses and grading them. Entity Tagging : Labeling the different entities based on their natureAudio Labeling: Have to identify the different entities in audio and tag them by following guidelines, examples like contacts, music artists, music albums, relationships, locations, point of interests. Query Classification : Classifying the Siri queries into the right segment depending on the nature of the request.Successfully met the quality standards in all projects worked without any remark in the quality.As part of the pilot team, participated in the complete overhaul of guidelines, adhering to the company’s privacy policies.The role involves active listening, transcribing audio files, and evaluating Siri’s response and language usage.Evaluating responses as per guidelines using gained cultural knowledge and analytical skills to provide detailed insights for Annotation.Demonstrated quick learning ability to effectively utilize various Data Annotation Tools.Provided enhancement ideas which will help the Annotation process better.Successfully met targets and monthly metrics.

Education

• Vignan's University

  Bachelor of Technology - BTech

  2009 — 2016