At Verity Labs, I build benchmark question sets from real-world data and use them to assess large language model responses for accuracy, relevance, and alignment. As Reviewer & Operations Manager, I review and approve annotation and evaluation work before delivery.
I started in Verity Labs’ video department, polishing in-game visuals for Anu acon's AI story game Whispers from the Star, before transitioning to LLM evaluation. I also led a team of data trainers.
At Invisible Technologies, I trained multimodal AI models using human-guided annotation and captioning across text, image, and video inputs. I applied SFT and RLHF to improve model behaviour and recognition accuracy.
