Sharavan Mudaliar

Utilize RLHF to evaluate multi-modal data (text, image, audio) against strict safety rubrics, identifying bias, hallucinations, and harmful content to optimize LLM grounding. Execute complex dialogue evaluation and prompt-response ranking via the MultiTango platform, making logical decisions on ambiguous edge cases to generate high-fidelity ground truth data. Perform root-cause analysis on error patterns and collaborate with governance leads to update Standard Operating Procedures (SOPs) and safety guidelines.

Analyst — AI/LLM Practice at Innodata India (2026-12 – Present)

• Utilize RLHF to evaluate multi-modal data (text, image, audio) against strict safety rubrics, identifying bias, hallucinations, and harmful content to optimize LLM grounding.
• Execute complex dialogue evaluation and prompt-response ranking via the MultiTango platform, making logical decisions on ambiguous edge cases to generate high-fidelity ground truth data.
• Perform root-cause analysis on error patterns and collaborate with governance leads to update Standard Operating Procedures (SOPs) and safety guidelines.

Data Annotator at Innodata India (2025-11 – 2025-12)

• Executed high-volume text categorization, image labeling, and search relevance tasks with 95% quality standard compliance, ensuring clean data collection.
• Conducted fact-checking and data auditing using internal tools, strictly adhering to dynamic instructions, SLAs, and data confidentiality protocols.
• Consistently met productivity baselines while making sound judgments on ambiguous information, raising execution queries proactively to resolve operational bottlenecks.

Master of Computer Applications in MCA – SIES College of Management Studies (SIESCOMS) (2024-08 – 2026-06)

Bachelor of Science in Computer Science – SIES College of Arts, Science and Commerce (2021-08 – 2024-05)