Robin L.
AI Tutor & Chemistry/Biology Specialist
Experience
Jan 2025 - Present
11 monthsAI Tutor & Chemistry/Biology Specialist
Mindrift
- Authored prompts, rubrics, and gold items for math, physics, computer science, chemistry, and biology; emphasized step-by-step solutions, unit handling, error explanations, and uncertainty language—resulting in higher reviewer agreement and fewer formatting defects.
- Reviewed model outputs for correctness, reasoning depth, and safety alignment; built calibration sets, difficulty bands, and decision trees for edge cases (e.g., oxidation-state slips, invalid resonance, disallowed clinical advice).
- Generated biology-focused prompts across cell biology, physiology, microbiology, and genetics; grounded responses with PubMed and NCBI citations and red-team checks for dual-use and ethical risks, enhancing factuality and tone control.
- Produced structure-spectra pairs and triage checks (unit gates, significant-figure policy, pH/pKa flags) adopted by peers; localized English to Swedish content for research and health summaries with precise terminology and register.
Jan 2024 - Present
1 year 11 monthsAI Linguistic Trainer – Swedish NLP & Multimodal Labeling
Alignerr & RWS (TrainAI)
- Authored 2,000+ complex, QA-verified prompt samples involving cognitive science and behavioral logic; passed biology domain assessment to support interdisciplinary annotation projects.
- Labeled complex language data with emphasis on intent clarity, psychological context, and multimodal scenarios (texts and images); delivered QA escalation reports contributing to edge-case protocol updates.
- Developed prompts using psychological framing to surface bias, ambiguity, and implicit flaws; validated intent recognition in user-facing workflows.
Jan 2023 - Present
2 years 11 monthsSwedish Language Content Creator (Chaya Project)
TransPerfect / DataForce
- Delivered linguistically and culturally localized content in Swedish and German; contributed to realism assessments for targeted copy, ensuring semantic and regulatory alignment.
- Reviewed AI-generated Swedish outputs for correctness and clarity; maintained 98.85% QA score across final two batches (813 tasks).
Nov 2022 - Present
3 years 1 monthFreelance AI Safety & Systems Consultant
Independent Contractor (DataAnnotation, Outlier AI, Appen, RWS TrainAI)
- Delivered 2,000+ high-precision evaluations of prompt completions, safety flags, and model behaviors; authored and adversarial-tested prompts to stress hallucination, misalignment, bias, and low-salience failures, achieving 98.55% QA on 813 Swedish tasks at TransPerfect.
- Specialized in rating and editing toxic materials at DataAnnotation: assessed explicit and graphic texts, images, and adversarial descriptions; localized toxic English content to Swedish while preserving intent, enforcing cultural sensitivity, and applying minimal-edit corrections for compliance.
- Created 20+ reusable LLM test scenarios simulating research, procedural, and reasoning workflows; included gold outputs, success and failure rubrics, edge-case variants, and regression harnesses to isolate specific failure classes.
- Collaborated with platform engineers and policy specialists to refine schema clarity, prompt scaffolding, and grading criteria; audited agent traces for ambiguity triggers, recalibrating scoring logic across biosciences, argumentation, and tool-use domains.
- Led reviewer calibration and inter-rater reliability efforts; developed concise playbooks, disagreement drills, and rationale logs that raised first-review pass rates and reduced second-pass edits on chemistry and STEM queues.
Jan 2021 - Present
4 years 11 monthsFounder & Governance Architect
Numineus Project
- Architected agentic logic frameworks and regulatory reasoning engines (e.g., Numineus Chem) for high-stakes domains (chemistry, law), normalizing heterogeneous inputs into traceable, audit-ready outputs—enabling repeatable public-sector AI deployments.
- Translated EU regulatory texts (GDPR, AI Act) into modular QA systems using logic gates, semantic triggers, and compliance checklists; produced governance playbooks, transparency notices, and human oversight patterns to meet risk classification thresholds.
- Designed RAG-style retrieval workflows and rapid prototypes with LangChain/OpenAI APIs.
- Built structured evaluation harnesses and defect taxonomies to expose failure modes pre-pilot; authored reference architectures, decision memos, and end-to-end documentation (technical/user) for delivery teams.
Jan 2018 - Present
7 years 11 monthsStockholm, Sweden
Independent Translator & LQA Reviewer
Independent Translator & LQA Reviewer
- Localized research communication and consumer health content from English to Swedish; enforced terminology control and register for scientific summaries.
- Performed quality reviews of life-science captions, summaries, and FAQs for tone, cultural fit, and layperson comprehension.
Summary
AI Governance Architect and LLM Reasoning Specialist with over three years of experience designing high-fidelity prompts, evaluation rubrics, and agentic workflows for regulated environments. Expertise in translating complex EU regulations (GDPR, AI Act) into auditable systems, ensuring privacy-preserving AI solutions while achieving 98.85% QA accuracy across 815+ audited tasks.
Proven in multilingual (Swedish/English) content evaluation, adversarial testing for safety and bias, and cross-domain applications (science/technology, legal principles, health, chemistry, biology). Hands-on leader in remote, fast-paced settings, delivering scalable documentation, inter-rater calibration, and error taxonomies that reduce rework and enhance model alignment. Passionate about bridging research with practical impact to foster ethical, interpretable AI.
Languages
Swedish
NativeEnglish
AdvancedEducation
Uppsala University
Law Studies (Public, Constitutional & EU Law) · Uppsala, Sweden · AB
Stensund Folkhögskola
Behavioral Science · Sweden · Top 5% of Class
Need a freelancer? Find your match in seconds.
Try FRATCH GPT More actions
Similar Freelancers
Discover other experts with similar qualifications and experience
- English
- Deutsch
2025 © FRATCH.IO GmbH. All rights reserved.