Mehwish Aziz — Product designer who builds what she designs
NVIDIA BioNeMo at IFF: A Protein Workspace for Scientists
Client: IFF, through TCS. Role: Only designer. Status: Shipped.
TCS AI Studio took AI ideas from feasibility to delivery across R&D, supply chain and IT. Protein Builder delivered conversational molecular development.
Plant proteins and fermentation require repeated lab work. Scientists needed BioNeMo without coding or operating each model separately.
Research audited web interfaces, hosted APIs and downloadable containers. Five personas: students, computational biologists, drug discovery scientists, model researchers and pipeline engineers.
Protein Builder runs in the browser: select ESMFold, AlphaFold or ProteinMPNN, paste a sequence or upload files, inspect the labelled predicted 3D structure beside the conversation, zoom, expand or download.
Scientists pick the model and stay the judge. Corrections feed back into training; nothing counts until wet-lab validation confirms it. Empty, generating, multi-upload and file-error states mapped to Streamlit components.
Closing the loop is a retrospective refined concept: residue confidence, flagging regions, attaching lab evidence, and reviewed flags reaching fine-tuning. Visuals are design mockups, not shipped IFF or NVIDIA screens.
Scientists reviewed as experts and a small participant group usability-tested before build. Shipped to IFF scientists as part of the AI Studio.
Get in touch
Email: mehwishaziz2025@gmail.com · LinkedIn · Download resume (PDF)