[agents] Automated Product Design: MSc, PhD, Post-Doctoral Funding & Internships: University of Cape Town

[Geoff Nitschke]

Several MSc, PhD, post-doctoral bursaries and research assistantship
(2022-2025) are currently available at the Evolutionary Machine
Learning Group, Department of Computer Science, University of Cape
Town (UCT), for the research topic: Multi-objective Evolutionary
Product Design.

Master’s (MSc) bursaries are fully funded for two (2) years, where
doctoral (PhD) and post-doctoral bursaries are funded for up-to three
(3) years.  One (1) to six month (6) long research assistantships are
also available.  All candidates will be physically based at the
Evolutionary Machine Learning Group, UCT, Cape Town, where the
research is done in collaboration with colleagues at Smarter Sorting,
Austin, Texas, USA.  For the purposes of the research assistantships,
working online from remote locations will also be considered.

Experience and expertise in Python and related machine-learning
libraries (SciPy, NumPy) and natural language processing frameworks
(SpaCy) is required, as is some implementation expertise and knowledge
of evolutionary algorithms and meta-heuristics. Some experience or
educational background in computational chemistry would be an
advantage but is not required.   Candidates will be working with
various publicly available data-sets including PubChem and ECHA, as
well as open-source data-bases such as PostgreSQL, data-cloud services
such as Snowflake, and data-analytics engines such as Elasticsearch.
Some experience working with OpenAI GPT2, Wolfram Mathematica and
Wolfram Alpha would be an advantage but is not necessary.

Funding amounts depend upon the position level and is commensurate
with the applicant’s skills and experience.  Applications will be
evaluated on a first-come-first-serve basis, and will continue to be
received and reviewed until the positions are filled.

Automated Product Design:

Currently consumer products ranging from plastics, to cosmetics to
batteries are manufactured from complex compositions of chemical
elements and pre-designed chemical substances.

An ongoing challenge in the design of new products is to: 1) minimise
manufacturing and material costs, 2) minimise the expected
environmental impact of the product (during manufacturing and after
consumer use), and to 3) only use specific (regulated) materials and
chemical compositions.  Thus, novel product design can be formulated
as a multi-objective optimisation problem.  Specifically, where the
design of any new product simultaneously satisfies all these
constraints, but the product designer is able to manage the weighting
(relative importance) of each design objective (constraint).  This
enables the automated production of a broad array of new products that
satisfy the design objectives to varying degrees.  A product designer
would then ideally select one of several automatically designed
products according their own specific constraints for how expensive a
product can be, what materials can be used in its manufacture, and
what the extent of its expected environmental impact can be.

This research investigates multi-objective evolutionary algorithms to
automate the design of a vast array of products, given a pre-defined
set of materials and chemical substances usable in the design and
manufacturing processes, and metrics for expected economic and
environmental cost.

For more information contact: Geoff Nitschke (gnitschke at cs.uct.ac.za).

-- 

Geoff Nitschke

gnitschke at cs.uct.ac.za | www.nitschke-lab.uct.ac.za

Associate Professor | Undergraduate Student Advisor
Department of Computer Science
University of Cape Town
South Africa