Prasenjit M.-AI Evaluation Specialist | Operations & Quality Professional | Search Quality Evaluator

AI Safety Analysis:

• Reviewed AI-generated responses for harmful, biased, and policy-violating content — applying detailed safety rubrics to flag and escalate unsafe model outputs.
• Conducted adversarial prompt testing and red-teaming to surface model failure modes, delivering structured observations to safety feedback pipelines.
• Provided RLHF safety ratings and preference labels to guide alignment and fine-tuning of large language models.

Search & Relevance Evaluation:

• Rated search results and AI responses for relevance, accuracy, and alignment with user intent across diverse query categories and domains.
• Applied E-E-A-T quality signals (Experience, Expertise, Authoritativeness, Trustworthiness) to assess content credibility and factual depth.
• Maintained consistent inter-rater reliability across high-volume rating batches, meeting strict accuracy targets and SLA deadlines.

Data Quality Auditing & Training Annotation:

• Audited AI output datasets for accuracy, consistency, and labeling errors — detecting mismatches and systematic quality issues prior to model training submission.
• Completed high-volume NLP labeling tasks: text classification, named entity recognition, intent tagging, and sentiment annotation across multi-domain datasets.
• Provided RLHF preference rankings and instruction-following quality scores across Q&A, summarization, conversational AI, and code evaluation tasks.
• Conducted QA reviews on peer-annotated batches, delivering calibration feedback to align team standards and reduce error rates.

• Managed structured workflows, executive communications, and reporting with high accuracy — reinforcing consistency and attention to detail critical for annotation and QA roles.
• Coordinated cross-departmental workflows and documentation to support timely decision-making and business continuity.
• Introduced process improvements that reduced administrative backlogs and improved overall team productivity.