


I build things that help us do science.


I’ll be at #S26MRS this week to talk about the work we’ve been doing on industrially-relevant materials discovery at @DuniaInnovation ! Reach out if you want to grab a coffee and talk about autonomous labs, agentic workflows, the sim2real gap, or job opportunities. Our presentation is on Thursday in session MT01.08.07 eppro01.ativ.me/appinfo.php?…

We’re hiring — reach out. Main reason I work at Dunia is our thesis — work backwards from the mess of deploying materials in industry to find where AI/automation fits, then push hard. We’ve been building for 3 years already. We must make the jump from the screen to the world.
If you’re looking for a mission that can change the world, join us. Catalysts are just the beginning. Semiconductors, batteries, composites, clean, abundant energy is the goal. We are entering the Dunia Age. Say hello@dunia.ai Join us -> dunia.ai

For those interested, here is the paper and the GitHub repo. Happy building! pubs.rsc.org/en/content/arti… github.com/fenning-research-…

The true metric for success here is not number of samples, but accelerated cycles of learning — a truly lagging indicator. Keep an eye on the Fenning lab at UCSD to see the fruits of PASCAL’s labor in the pipeline.

I am personally a huge advocate for autonomous experiments, but in my experience these workflows shine best in large-scale and repetitive experimental campaigns that are uncommon in university labs.

While capable of “self-driving” using Bayesian Optimization, >95% of samples were designed by human operators. The art of coaxing the machine too often gets in the way of the science — especially when experimental goals change too quickly to justify “getting it right”.

PASCAL was “down” for 1/3 of the time, with average ~2 months of operation between failures. PASCAL had ~10 trained operators from grad students to postdocs. (Note I’m assuming that any >2 week idle period is a system failure. This is an overestimate — people take vacations!)

In a typical (median) week, PASCAL made 29 samples. Sample throughput ebbs and flows with the demands of ongoing projects in the lab; we made over 100 samples in 1/3 of the operational weeks!

PASCAL is designed to make and measure perovskite solar cells. One “sample” could be a 30 minute heating test, or a complete solar cell. In its best 24 hour period, PASCAL performed 400 samples -- guess which kind that was 😅youtube.com/watch?v=dKTjTsvg…

I built a robotic lab in grad school. Last week it made its 10,000th sample. Here's some stats from three years of operation.

We're hiring! We are looking for a postdoctoral researcher in energy materials and lab automation. Join us at the beautiful #Seattle!
The UW Sun Lab is looking for a postdoctoral researcher that have some experience in material science or chemistry. Please send @SunShijing a message if you are interested.

McMaster-Carr, the gold standard of web design. They didn't have anyone to look up to, there is nothing like it. They didn't use a design framework. They focused on fundamentals of design, not fads and fashion. mcmaster.com/ They didn't have to do it, but they did.

Data-driven materials folks — how are you storing all your data? Especially in the context of automated experiments. Files and folders? Bespoke relational databases? Anyone using GEMD from Citrine? Would love to discuss pain points!

When groups meet to brainstorm, good ideas are lost. People bite their tongues due to conformity pressure, noise, and ego threat. A better approach is brainwriting: generate ideas separately, then meet to assess and refine. Group wisdom begins with individual creativity.
