At OneForma CrowdGen, Trustscale.ai, Aligners Outlier & Telus, I’ve evaluated LLM responses and designed rubrics across Lighthouse 3 and Isaac pipelines. I maintained 96%-100% benchmark accuracy across multi-tier evaluation queues.
I reviewed Search, Maps, and UX quality across Whatcom, Milky Way, and Hydra pipelines. I also validated Hindi and English audio transcription and speech quality for Project Jigglypuff and A13372.
Using Python, Pandas, and spreadsheets, I cleaned, structured, and validated datasets with 98%+ accuracy. My work also includes safety benchmarks, adversarial prompts, and factual consistency reviews.
Through Zooniverse’s Snapshot Wisconsin project, I contributed image classification and tracking data. I’ve also compiled investigative reports on administrative transparency and conservation using field tracking and analytical observation.
