Tannu Rajput

Proof of Experience

Srujanee · May 2026 – Present

Reviewed and quality-checked speech/audio transcriptions across diverse datasets for accuracy and consistency.
Performed quality checks on transcriptions to ensure linguistic accuracy and consistency.
Followed transcription guidelines and formatting standards to maintain quality benchmarks.
Identified and corrected errors to improve overall transcription accuracy and reliability.

RWS · Jul 2026 – Sep 2026

Evaluated model-generated Text-to-Speech (TTS) audio for pronunciation, naturalness, intonation, fluency, and overall audio quality.
Conducted real-time conversations with AI chatbots and evaluated their responses for accuracy, naturalness, relevance, and conversational quality.
Provided structured feedback and quality ratings to support model fine-tuning and improvement.
Followed project-specific evaluation guidelines to ensure consistent and reliable assessment.

Transsion · Nov 2025 – Apr 2026

Reviewed and quality-checked transcripts generated from users' everyday voice queries to AI voice assistants and conversational systems.
Identified transcription errors, inconsistencies, omissions, and misinterpretations in speech-to-text data.
Ensured transcripts accurately represented the user's original spoken input while preserving context and intent.
Handled a wide variety of real-world voice queries, including questions, commands, conversational requests, and everyday interactions.

Krutrim (Ola) · Mar 2025 – Feb 2026

Conducted quality assurance for annotations, translations, and linguistic tasks.
Localized prompts and edited responses for the DPO dataset following strict accuracy and safety guidelines.
Evaluated LLM-generated Hindi outputs based on fluency, factual accuracy, and cultural relevance.
Performed ASR transcription for diverse audio datasets with high accuracy and conducted quality checks on ASR transcripts to ensure compliance with linguistic and formatting guidelines.
Performed quality validation on synthetic textbook content created by LLMs in Indic languages.
Collaborated on bias mitigation to build neutral and inclusive AI datasets.
Evaluated and corrected AI-generated Hindi-English translations.
Supported hiring processes, conducted accuracy audits, and contributed to annotation guideline development.
Edited and refined machine-generated Text-to-Speech (TTS) scripts to achieve natural, fluent, and human-like speech output.
Improved grammar, syntax, punctuation, tone, and contextual accuracy across large-scale multilingual datasets.
Performed quality assurance of TTS scripts to ensure linguistic accuracy, naturalness, and prosodic consistency.
Designed and generated high-quality image prompts to train AI models for accurate visual outputs.
Performed quality checks on image outputs, ensuring adherence to project guidelines.

Krutrim (Ola) · Dec 2023 – Mar 2025

Contributed to improving AI model outputs by fine-tuning datasets and applying Reinforcement Learning from Human Feedback (RLHF).
Performed comparative analysis of AI-generated text to evaluate coherence, accuracy, and contextual relevance.
Benchmarked company's AI models against industry standards, ensuring optimal performance in multilingual NLP tasks.
Identified linguistic biases, errors and inconsistencies in AI-generated text to enhance model fairness and accuracy.
Designed domain-specific prompt-response datasets (education, healthcare, social media).
Created structured datasets for AI model training, improving language diversity and cultural accuracy.

No completed sessions yet. Anything Tannu finishes shows up here with the evidence behind it.