Saharsh P.

AI Trainer Evaluator | Transcriber | LLM | Data Annotator | QA | RLHF

Law and Commerce graduate and Semi Qualified Company Secretary (CS Executive cleared) with over six years of hands on experience in legal drafting and corporate compliance, along with more than ten years of experience as an educator. Since January 2025, actively working as an AI Trainer and Evaluator, specializing in AI training, LLM evaluation, transcription, function calling, and data quality enhancement. Former Legal Trainee at law firms with substantial exposure to corporate legal documentation, regulatory compliance, and legal advisory functions. Currently engaged as a Generalist AI Trainer / Evaluator with leading global AI platforms including Labelbox, Outlier.ai, Alignerr, and OneForma. Work contributions include supervised fine tuning (SFT), prompt evaluation, audio and text transcription, legal domain task development, function calling and tool use evaluation, and rigorous assessment of AI generated outputs for accuracy, reasoning quality, and instruction adherence. I have contributed to multiple projects involving data annotation, data labeling, AI model evaluation, reasoning validation, RLHF style comparisons, and quality assurance for agent based and LLM driven systems. My responsibilities include transforming structured inputs into realistic task scenarios, evaluating agent reasoning paths, comparing multiple model outputs, verifying pass/fail outcomes, and reviewing automated QA decisions. I also document clear solution outlines and reasoning justifications to ensure fair, consistent, and explainable evaluations. I am experienced in working with structured data formats such as JSON and YAML, test case creation, scenario based and agentic evaluations, and tasks requiring strong logical reasoning, high attention to detail, and strict compliance with system instructions and evaluation guidelines. Alongside AI work, I am the Founder and Educator at Gurukul Tutorials, Ranchi . I teach Science, Mathematics, English, Hindi, Accountancy, and Economics to high school students, and core foundation level subjects to CA, CS, and CMA aspirants. I have extensive experience in evaluating written responses, providing structured, criteria based feedback, and designing analytical assignments skills that directly translate to AI task creation, annotation, and evaluation workflows. Key Strengths & Skills 1. AI Training & LLM Evaluation – Model output assessment, reasoning validation, RLHF comparisons, and QA. 2. Data Annotation & Data Labeling – Text, classification, reasoning, and scenario based tasks. 3. Function Calling & Tool Use Evaluation – Instruction adherence and agent behavior analysis. 4. Transcription – High accuracy audio and text transcription. 5. Prompt Engineering & Prompt Evaluation – Prompt creation, response checking, and consistency analysis. 6. JSON & YAML – Structured data handling and task definition. 7. Proofreading & Editing – Accuracy, clarity, grammar, and coherence. 8. Translation – Context aware and culturally accurate translation. 9. Legal Drafting – Corporate and legal documentation expertise. 10. Quality Assurance – Automated QA review and evaluation consistency checks.