I'm Wei Sheng, a Year 4 Computer Science undergraduate at the National University of Singapore, currently interning at Ren building production AI grading infrastructure.

Some things I've built:

• An end-to-end AI essay grading pipeline in Python - GPT Vision extraction, dual-method grading strategy, per-rubric scoring with justifications
• Evaluation infrastructure to detect hallucination in AI grading outputs using embedding cosine similarity and an LLM judge against gold-standard teacher annotations, methodology wrote in blog here
• A concurrent load testing framework for the grading API, instrumenting LLM token costs, latency, and memory across parallel jobs to model cost-per-submission

Full breakdown of my work at Ren - lws49/work