Materials Scientist and PhD researcher evaluating LLM reasoning, scientific accuracy, and RLHF workflows across Physics, Chemistry, and Mechanical Engineering — turning deep STEM expertise into higher-quality AI.
Irfan Khan · Sydney, NSW
Assessing reasoning, factual correctness, and completeness in AI outputs across materials science, thermodynamics, and semiconductor physics.
Building expert annotations, evaluation rubrics, and reasoning benchmarks that lift model performance through human-feedback workflows.
Surfacing factual errors, reasoning gaps, and inconsistencies in complex STEM datasets to protect dataset integrity.
Writing high-quality prompts and gold-standard answers that diversify training data and reduce model bias.
Photovoltaics, nanotechnology, spectroscopy, and semiconductor materials backed by a decade of hands-on research.
Translating dense engineering into precise, clear explanations suited to instructional and evaluation datasets.
Open to AI training, STEM evaluation, and RLHF roles. Reach out to talk about how deep domain expertise can raise your model's scientific reasoning.