Kanishk Surwade

AI & LLM Analyst with hands-on experience evaluating, annotating, and quality-validating multimodal datasets (text, image, document, audio, video) for Generative AI and LLM systems in a live, confidential production environment. Skilled in LLM output review, prompt and response evaluation, AI annotation, and data quality analysis, with proven ability to maintain precision and consistency across large-scale datasets while meeting strict SOPs and productivity benchmarks. Experienced in supporting AI model fine-tuning and building quality-monitoring dashboards on Google Cloud Platform.

AI & LLM Analyst at Innodata Inc. (2025-11 – Present)

• Performed annotation, evaluation, and quality analysis on multimodal datasets (text, image, document) for Generative AI and LLM systems in a live, confidential production environment.
• Conducted LLM output review and prompt analysis, applying structured QA frameworks to validate model responses against accuracy and consistency benchmarks.
• Executed bounding box annotation, object labelling, and image segmentation on large-scale visual datasets, maintaining high precision across thousands of samples under strict client SOPs.
• Collaborated with QA leads during calibration sessions to resolve annotation ambiguities and edge cases, improving inter-annotator consistency.
• Supported AI model fine-tuning through systematic data cleaning and quality validation; built Power BI dashboards on GCP to monitor annotation throughput and quality metrics.

Data Analytics Intern at Deloitte (2025-09 – 2025-09)

• Analysed manufacturing and operational datasets to identify downtime and performance patterns using interactive dashboards.
• Created calculated metrics and KPIs to visualize trends across systems and categories.
• Performed structured analysis in Excel to support fairness and classification use cases.
• Delivered insights that improved monitoring, reporting, and decision-making processes.

B.Tech in Automation and Robotics – JSPM Rajarshi Shahu College of Engineering (2021 – 2025)