Kanishk Surwade

AI & LLM Analyst with production experience in multimodal AI evaluation, LLM output quality analysis, prompt engineering, hallucination detection, grounding assessment, and RLHF-aligned annotation on confidential enterprise datasets. Skilled in side-by-side model evaluation, multi-turn conversation testing, SOP-driven quality benchmarking, and cross-functional collaboration. Experienced with Google Cloud Platform, Power BI, and data operations in high-volume remote AI workflows.

AI & LLM Analyst - Innodata Inc. - Noida, India (Remote)

(2025-12 - 2026-07)

• Conducted LLM evaluation including hallucination detection, grounding verification, side-by-side model comparison, and multi-turn conversation quality assessment on confidential production datasets.
• Performed prompt analysis and prompt engineering to assess instruction-following, response coherence, and output format compliance across generative AI systems.
• Executed multimodal AI annotation and quality assurance (text, audio, video, image) following strict SOPs to maintain inter-annotator accuracy and consistency standards.
• Supported AI fine-tuning workflows through data cleaning, quality validation, and structured feedback aligned with RLHF best practices.
• Built Power BI dashboards on GCP to monitor operational throughput, flag quality deviations, and drive data-driven process improvements.
• Escalated edge cases and SOP ambiguities during calibration sessions, coordinating with cross-functional leads to maintain downstream data integrity.

Data Analytics Intern - Deloitte - Virtual

(2025-09 - 2025-09)

• Analysed manufacturing and operational datasets to identify performance gaps; developed KPI dashboards and calculated metrics in Excel to improve reporting and decision-making.

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