AI-driven sustainable material design & development - ON-1107

Project type: Research
Desired discipline(s): Engineering - other, Engineering, Computer science, Mathematical Sciences, Chemistry, Natural Sciences
Company: erthos Inc.
Project Length: 6 months to 1 year
Preferred start date: As soon as possible.
Language requirement: English
Location(s): Mississauga, ON, Canada
No. of positions: 3
Desired education level: PhDPostdoctoral fellowRecent graduate
Open to applicants registered at an institution outside of Canada: No

About the company: 

As a climate technology company specializing in sustainable material design, erthos® focuses on rapidly reducing the global dependency on plastics by accelerating the scale and widespread adoption of biomaterials. We develop bespoke sustainable materials (erthos® Ingredients) and digital tools (ZYA™) tailored to meet our clients’ sustainability and functional goals.

Describe the project.: 

This project supports the development of ZYA™, an AI-powered platform that will help scientists and packaging engineers manage packaging innovation projects and rapidly design, test, and optimize compostable, biobased materials. It replaces slow, manual experimental methods with intelligent, data-driven workflows to accelerate product development.

ZYA’s users include R&D scientists at consumer goods, packaging, and chemical companies. The platform supports three core functions:
1. Ingredient Intelligence – Access a structured database of sustainable inputs.
2. Formulation Generation & Predictive Testing – Create material recipes based on mechanical, processing, and sustainability targets. Simulate and optimize material performance before lab trials.
3. End-to-end project management - Create and track packaging innovation projects involving cross-functional teams

ZYA is being piloted by global CPG. The market includes the $600B global plastics industry and the growing $30B+ bioplastics sector, which faces significant R&D and scale-up challenges.

Research outcomes will include predictive property models, ingredient-performance mappings, and reproducible digital workflows for material innovation. Candidates will support data collection/validation, development and testing of machine learning models, and deployment of models to our front-end application.

Required expertise/skills: 

We are looking for two distinct roles:
1. Computational materials scientist - ideally with a PhD in Machine Learning, Computational Sciences, Chemistry, or Physics. Core requirements:
● Experience applying ML to materials design, drug discovery, or computational chemistry/physics
● Proficiency in molecular representations (e.g., SMILES, molecular graphs, 3D structures); hands-on with RDKit, DeepChem
● Strong Python skills; experience with PyTorch, TensorFlow, Scikit-learn
● Familiarity with graph neural networks (GNNs), Transformer models, and classical ML (e.g., XGBoost)
● Knowledge of uncertainty quantification, Bayesian optimization, and scientific ML workflows
● Demonstrated scientific contributions (e.g., publications, open-source/GitHub)
2. Machine learning, NLP/LLM - ideally with a Master’s or PhD in Machine Learning, Computer Science, or a related field. Core requirements:
● Experience developing, fine-tuning, or applying LLMs (e.g., GPT, BERT, T5) to scientific or technical domains
● Proficient in Python and ML frameworks (e.g., PyTorch, Hugging Face Transformers)
● Strong understanding of tokenization, attention mechanisms, prompt engineering, and model evaluation
● Familiarity with transfer learning, retrieval-augmented generation (RAG), and instruction tuning
● Demonstrated research contributions (e.g., publications, open-source/GitHub)