Vikash ajmera User

Review queries and responses for guideline compliance and safety, assess intent and readability, and perform actuality evaluations to identify misinformation and improve response quality.

Design and optimize prompts and evaluate LLM responses to improve response quality and multilingual alignment; authored adversarial prompts and evaluation reports to identify model weaknesses.

Evaluate LLM outputs for effectiveness and cultural relevance, providing actionable feedback and reports to improve multilingual model performance and guideline adherence.

Created precise multilingual datasets and performed speaker diarization for projects like Fireweed and Plumeria V2, ensuring contextual relevance and high-quality training data for LLMs.

Designed AI evaluation tasks and prompts for financial modeling, risk assessment, and market forecasting; evaluated outputs for factual accuracy and compliance with US financial standards.

Analyzed and refined AI-generated responses for accuracy, coherence, and contextual relevance; conducted red-teaming and benchmarking to expose hallucinations and biases.

Conduct linguistic QA and edit Hindi AI outputs for naturalness and cultural appropriateness; designed role-plays and evaluated bilingual datasets to improve communication standards.

Edited promotional and social media videos using Premiere Pro and After Effects, applying color grading and audio mixing to deliver optimized short- and long-form content on deadline.

Taught mathematics to primary and secondary students, designed differentiated lesson plans, and provided personalized support to enhance student learning outcomes.

Completed a Bachelor of Commerce (B.Com) at Gauhati University, focusing on core commerce subjects and coursework from 2006 to 2009.