Ipek Platt

Highly competent analyst with a background spanning AI and LLM evaluation, engineering, and law. Proven track record in managing data quality standards, regulating worker output, and operating reliably within remote, asynchronous environments requiring high precision and consistency.

AI Red-Team (Freelance) - Handshake AI - Remote

(2026-07 - 2026-08)

• Engineered complex adversarial prompts to test model boundaries, evaluating behavioural alignment, safety guardrails, and compliance under stress.
• Designed systematic jailbreak payloads (including linguistic obfuscation and role-play overrides) to uncover vulnerabilities to prompt injection and data leakage.

AI Evaluation & Reasoning Specialist (Freelance) - Data Annotation Tech - Remote

(2026-04)

• Evaluated AI-generated and LLM textual and multimodal outputs for factual accuracy, reasoning quality, instruction adherence, formatting compliance, completeness, conciseness, safety alignment, and overall response reliability.
• Conducted mathematical reasoning evaluation across advanced A-Level topics including algebra, calculus, mechanics, geometry, and statistics.
• Reviewed coding-related outputs involving debugging, syntax validation, algorithmic reasoning, and software explanation tasks.
• Performed individual and comparative evaluation of model outputs using scoring frameworks, providing robust justification for scores.
• Designed and refined prompts to assess reasoning behaviour, instruction following, robustness, and failure modes across diverse task categories.
• Developed adversarial and edge-case prompts to identify instruction following weaknesses, hallucinations, logical inconsistencies, policy weaknesses, and robustness limitations in AI systems.
• Investigated factual claims using multiple sources, verified information accuracy, and documented findings through structured written assessments.
• Created and applied detailed rubrics to regulate LLM output, maintaining or improve response quality.
• Provided generalist and STEM colleagues with feedback and corrected data outputs to tight deadlines.
• Delivered structured feedback to support AI alignment, safety compliance, and model performance improvement.
• Operated within NDA-protected workflows with ongoing task allocation, maintaining sustained compliance with internal quality thresholds required for continued participation.
• Worked effectively in large-scale asynchronous annotation environments requiring consistency and attention to detail.
• Reviewed colleagues work to ensure accuracy.

LLB - Law - Birkbeck, University of London (2025 - 2026)

BEng - Civil Engineering - University of Warwick (2006 - 2007)

A-levels - Brentford School for Girls (1998 - 2006)

GCSEs - Brentford School for Girls (1998 - 2006)