Rabiul Alam Roni
← Back to all work

FoodProt & FoodProt AI

Two complementary research platforms for plant-protein functionality: a structured research-data platform and an AI-assisted evidence-extraction platform with human review.

Question
How can plant-protein functionality evidence from literature, experiments, and molecular dynamics be organized into reproducible, machine-learning-ready research data with provenance and human review?
Why it matters
Protein-functionality knowledge is scattered across sources in inconsistent formats. Comparable, provenance-preserving data makes cross-study analysis and reliable modeling possible.
My role
Research Platform Developer — scientific workflow design, structured data, testing, deployment, and maintenance. Python is my primary language; Next.js/TypeScript and PostgreSQL/Supabase are project-based experience.
Status
Both platforms are deployed and under active development.
Methods
  • Structured data entry, import, review, and quality control
  • Provenance with original and standardized units
  • Confidence, abstention, and human scientific review
  • Filtering, visualization, and machine-learning-ready export
Outputs

Overview

FoodProt and FoodProt AI are complementary research platforms developed in the Zhao Food Chemistry Lab. Together they turn heterogeneous plant-protein functionality evidence into structured, machine-learning-ready research data, with provenance and human review central throughout.

FoodProt

Deployed · in active development

A research-data platform that brings literature-derived measurements, collaborative experimental data, and molecular-dynamics metadata into one structured resource for comparison and reuse, with data entry and import, review, quality control, filtering, visualization, and machine-learning-ready export.

Visit FoodProt (opens in a new tab)

FoodProt AI

Deployed · extraction and review features in development

An AI-assisted literature and evidence-extraction platform. Model-generated information is treated as candidate evidence for scientific review, with experimental conditions, original and standardized units, provenance, confidence, and abstention recorded for each extracted value.

Conceptual workflow
  1. Literature discovery
  2. Study screening
  3. AI-assisted extraction
  4. Quality control
  5. Human scientific review
  6. Verified structured data
Visit FoodProt AI (opens in a new tab)
Abstract submitted · Poster requested

Rabiul Alam Roni, Changqi Liu, and Jing Zhao. (2027). FoodProt: Provenance-aware plant protein data integration for rapid analysis and machine learning.

  • ACS Spring 2027, New Orleans, LA — 2027
All publications & presentations