Jennifer Rose
Training Specialist @micro1
Signup · Get unlimited contacts
WORK HISTORY
Training Specialist @micro1
US
As an AI Trainer and Training Specialist, I design and deliver instructional content and feedback mechanisms to improve large language model performance. I work on data labeling, prompt-response evaluation, fine-tuning datasets, and creating training modules that enhance AI accuracy, coherence, and alignment with human values. My contributions help refine AI outputs for real-world applications and user-centric interactions.Skills & Keywords:AI Training | LLM Fine-Tuning | Data Labeling | Model Alignment | Instructional Design | Training Data Curation | Human Feedback Loops | AI Performance Optimization | Prompt-Response Evaluation | Machine Learning Support | AI Education | Quality Assurance
EDUCATION
The University of Akron
Bachelor's degree, B.A. in Anthropology & Philosophy (Interdisciplinary)
The University of Akron
Master's degree, M.A. in Geography (Spatial Data Focus)
ABOUT JENNIFER ROSE
AI Safety & Sociotechnical Risk Specialist | Red-Teaming & Adversarial Evaluation ExpertI turn a deep, lived understanding of high-risk human predation into actionable safeguards for digital environments. My focus is on extreme edge cases and adversarial manipulation, analyzing how malicious actors identify and exploit weaknesses in complex systems—whether social, legal, or technological. My work involves stress-testing not just for bias, but for the potential vectors of coordinated abuse, identifying critical failures in systems designed to protect the vulnerable.What I Do:Conduct structured red-teaming and safety evaluations for leading AI developers, probing for hidden risks, logical flaws, and ethical vulnerabilities.Design adversarial prompts and scenarios that test model reasoning, alignment, and guardrails.Analyze AI outputs for bias, cultural nuance, ethical gaps, and logical inconsistencies.Research the bidirectional impact of AI training—observing how models learn from humans, and how humans are meta-cognitively shaped by the work.My Approach Is Rooted In:Interdisciplinary Analysis: Blending anthropology, ethics, logic, and spatial-critical thinking.Forensic Pattern Recognition: Identifying how systems can be exploited by coordinated actors.Rapid Domain Acquisition: Mastering complex topics to stress-test AI knowledge and reasoning under pressure.I am driven by the mission to make AI safer, smarter, and more aligned with human values. Currently open to contract and project-based roles in AI safety, red-teaming, adversarial evaluation, and human-AI interaction research.Let\'s connect if you\'re building safer systems, advancing evaluator methodologies, or exploring the human side of AI training.Currently contributing to LinkedIn\'s Red Team Hard Classifier Evaluation project, where I generate adversarial examples and evaluate safety classifier performance on complex edge cases. This work involves creating high-signal test prompts across multiple harm categories, applying structured severity judgments, and developing policy-grounded rationales to identify classifier vulnerabilities including severity miscalibration, policy boundary confusion, and context blindness.
This profile is compiled from publicly available professional sources. Unifers is not affiliated with or endorsed by LinkedIn. Request removal of this profile.