Skip to content
Tushar Ranjan Panda

Tushar Ranjan Panda

Professional
Analyst - AI/LLM Practice at Innodata India Pvt. Ltd. Cuttack, 21, IN AI-First Professional
QR Code

💼 Work Experience

Analyst - AI LLM Practice

Innodata India Private Limited Contract

Nov 2025 - Jul 2026 • Noida, Uttar Pradesh, India (Remote) • Information Technology

Responsibilities

- Multimodal Data Annotation & Alignment: Executed high-precision annotation and RLHF/SFT evaluation
across text, image, video, and audio modalities to train and align next-generation multimodal LLMs for accuracy
and safety.
- Complex Prompt Engineering & Benchmarking: Designed and evaluated complex multi-turn prompt-response
pairs to stress-test reasoning, instruction-following, and edge-case capabilities of frontier AI models.
- Response Evaluation & Pairwise Ranking: Analyzed and ranked multi-model outputs based on helpfulness,
relevance, formatting, and factual consistency to directly inform model preferences.
- Cross-Modal Data Verification: Validated complex cross-modal alignment including text to image relevance,
audio transcription fidelity, and visual context modeling to ensure high data fidelity for domain-specific models.
- Model Safety & Hallucination Reduction: Evaluated model outputs to detect factual inconsistencies, bias, and
toxic content, providing structured feedback to minimize hallucinations and improve response safety.

Multimodal Annotation

Transcriptionist

LXT AI India Private Limited Contract

Oct 2023 - Feb 2025 • Hyderabad, Telangana, India (Onsite) • Information Technology

Responsibilities

- High-Precision Audio Transcription: Transcribed, formatted, and verified complex audio datasets into highaccuracy
text, strictly adhering to custom formatting and style guidelines.
- Acoustic & Phonetic Accuracy: Captured multi-speaker audio with exact attention to detail, accurately handling
varying accents, speech paces, background noise, and overlapping dialogue.
- Audio Annotation & Time-Stamping: Applied precise timestamping, noise tagging, and speaker identification
to structure unstructured audio files for downstream processing.
- Quality Control & Proofreading: Conducted thorough self-audits and proofreading of transcriptions to maintain
a high accuracy benchmark across all delivered audio batches.
- Production & Deadline Delivery: Consistently met daily audio-hour processing targets and quick turnaround
times while maintaining strict quality parameters.

Audio Transcription Audio Annotation

🎓 Education

Bachelor of Technology - Computer Science & Engineering

Biju Patnaik University of Technology, Odisha

2017 - 2021

Skills & Expertise

transcription 50%

About

Dedicated and results-oriented AI Professional with experience in AI data training, evaluation, and speech transcription.

Specialized in evaluating Large Language Model (LLM) responses and fine-tuning multimodal datasets spanning text,

audio, video, and image inputs. Proven track record in rigorous quality assurance, prompt-response auditing, and

maintaining high ground-truth data accuracy for industry-leading AI models. Experienced in executing Reinforcement

Learning from Human Feedback (RLHF) workflows, identifying model edge cases, and ensuring strict adherence to

complex guidelines.