Mara Cairo is keen about utilizing AI for good. She has a Bachelor of Science in Electrical Engineering from the University of Alberta and holds her P.Eng. and PMP designations. Before becoming a member of Amii, she labored within the {hardware} growth house, the place she helped shoppers take their merchandise to market, with a concentrate on micro and nano-fabrication.
As Product Owner of Advanced Technology at Amii, Mara leads a technical staff that helps {industry} companions construct machine studying capability inside their group by offering steering and experience to develop predictive fashions. Her staff works with shoppers who’re dedicated to advancing alongside the AI adoption spectrum by making use of machine studying to their most difficult enterprise issues.
Amii (Alberta Machine Intelligence Institute) is one in all Canada’s preeminent facilities for AI, they companion with firms of all sizes, throughout industries, to drive innovation technique and supply sensible steering and recommendation, company coaching and expertise recruitment companies.
We sat down for an interview on the annual 2023 Upper Bound convention on AI that’s held in Edmonton, AB and hosted by Amii.
What initially attracted you to electrical engineering?
As a child, I simply actually favored constructing issues. My mother would carry house a fan when it was scorching in summer season, and I’d wish to construct it. I bear in mind rising up as a teen, I had a mobile phone, a type of Nokia’s that you might take aside and I’d take it aside and put bejewels throughout it on the within and the antenna. But once I opened it up, it was like, “Holy crap, what’s in right here? What’s occurring?” It was actually attention-grabbing to me.
I at all times excelled in math. So, placing all of these collectively, my mother and father additionally pushed me within the engineering path as a result of I used to be good at math, I had only a common curiosity in electronics and wished to know extra about it, that is type of what drew me in to start with.
Also, in engineering, I simply actually favored the concept of making use of math to real-world issues. Yeah, okay, cool, math is nice and thrilling and enjoyable for me, however with engineering you possibly can apply it to resolve exhausting issues. It appeared type of the proper mesh of issues that might result in an attention-grabbing profession.
Your mother and father sounded very proactive in supporting your pursuits.
Yeah. My dad particularly. He says he noticed it in me from a younger age and simply at all times pushed me in that path. I used to be at a Women in AI occasion final evening too and we talked about eradicating some limitations and making it a extra approachable area for girls. And I did not actually see that as a barrier as a result of, once more, my mother and father had been like, “This is what it is best to do. It’s not a query of your gender or something. It’s simply it is a talent you could have. You ought to naturally type of observe it and nurture it.” I by no means felt prefer it wasn’t for me, which helped clearly.
Before becoming a member of Amii you labored within the {hardware} growth house to concentrate on micro and nanofabrication. Could you outline these phrases?
Definitely. So, in electrical engineering, I took the nanoengineering possibility. It was the specialty round designing and manufacturing on the micro and nanoscale. When we speak about a nanometer, we’re speaking a couple of millimeter divided in one million is a nanometer. A really, very small scale. And that is cool. These issues are so small you possibly can’t even see them with the bare eye. But I might take this specialization to learn to manufacture on that scale and design issues on that scale.
We dwell in a really linked world. There’s electronics throughout us and we want to have the ability to design electronics for the packaging and house constraints. We’re consistently making an attempt to make issues smaller and smaller. You take one thing cumbersome, a prototype, and also you want to have the ability to make it reproducible and scalable. Nanofabrication is absolutely in regards to the instruments and the methods that you just use to design and manufacture on that type of degree.
This is from manufacturing microchips to taking these two completely different chips and connecting them electrically to the ultimate packaging. Doing all of that on the microscale requires a distinct approach than constructing one thing on our human scale. The micro and nanofabrication are simply across the chemical processes that you just use and {the electrical} processes, the packaging that it is advisable be certain that these are hermetically sealed and protected against their atmosphere.
Outside of microchips, what could be one other utility or use case?
We labored on a whole lot of tasks like fiber optics. Again, all of it finally should come to some kind of processing unit that is taking in alerts or producing alerts. We did work within the telecom {industry}, optics, cameras, all of that stuff. But the brains of it are typically some kind of microchip within the center. But there’s additionally the sensors which can be feeding their alerts into no matter processing unit you are utilizing. So numerous manufacturing methods for constructing no matter kind of sensor or enter or output system that we want.
What are a few of the challenges behind engaged on one of these nanoscale?
One piece of mud can break your entire day. Things you are engaged on are the identical dimension because the mud within the air. So, you fabricate in a clear room. The clear room is absolutely an atmosphere that is defending what you are engaged on from you as a human, as a result of we’re very soiled as people, we’re consistently type of spitting out particulates, our garments are particulating, the make-up that we’re sporting it is making the air soiled. We have to get rid of as a lot of that as potential in order that the issues that we’re constructing are clear and clear of that kind of contaminant.
Another problem, there’s nice methods to construct these clear rooms and there is a entire type of examine and science behind that, however the different problem is taking it out of the lab as a result of finally these items are going for use in our very soiled world. That’s when the packaging turns into necessary. We nonetheless want to have the ability to entry these gadgets, however we have to do it in such a means that we’re not contaminating the atmosphere, the packaging. So hermetically sealing issues, ensuring it is fully sealed, nothing’s getting in or out. That’s one other set of challenges that I noticed. We would have one thing that works nice on a lab bench in a managed setting, however typically many of the issues that we’re constructing are supposed to be introduced out into our soiled world. That was difficult as properly.
Again, from manufacturing all the way in which to taking it to its remaining vacation spot, it is simply very particular type of issues and environmental issues if you’re coping with issues that small. Also, issues do not at all times behave as anticipated on that small of a scale. In our bodily world, we anticipate issues to work a sure means, however if you get all the way down to the micro and nanoscale, the bodily world turns into just a little bit completely different, and you may’t at all times anticipate the outcomes. That’s an entire different area of examine.
What could be some examples of being completely different than the common bodily world?
Passing present via a wire. We have our chargers and our telephones and we’re passing present via it. When you are passing present via a wire that is sized like a strand of hair, there’s clearly warmth issues and issues will simply begin behaving otherwise as a result of, once more, the house and the dimensions constraints.
What is your present function at Amii, and the way does your staff assist {industry} companions?
My present function at Amii is vastly completely different from the world of micro and nanotechnology.
I’m Product Owner of the Advanced Technology Team at Amii. I lead a staff of principally machine studying scientists and mission managers who’re all working with our completely different {industry} companions to resolve their enterprise issues via the applying of machine studying.
We’re very industry-focused, all about bridging the hole between what’s taking place in academia, the entire actually nice breakthroughs with machine studying and AI however making use of them to our {industry} companions greatest wants. We reply to these wants by primarily serving to our shoppers discover the talents and the experience that they want to have the ability to transfer the work ahead.
We run our internships and residencies program via the superior expertise staff. So, I’m hiring rather a lot. Recruitment shouldn’t be my background, nevertheless it’s one thing I do rather a lot now. And it is all about type of matchmaking, discovering the precise ML expertise to position on our consumer’s mission. We rent these of us as Amii staff for a set time period and provides them a whole lot of assist and mentorship, however actually, they’re devoted to work on the consumer’s mission and transfer that ahead. It’s a means for our shoppers to get entry to expertise with out having to do the recruitment themselves. Amii has some fairly good model recognition, we’re in a position to carry actually nice expertise in after which place them on these {industry} tasks.
A possible advantage of the system is the consumer having the chance to rent these of us after the time period with us is completed. We need this expertise to remain right here. We don’t need mind drain. We’re giving the consumer a little bit of a leg up in order that they’ll strive the expertise out, check out the mission, get a really feel for what machine studying truly is, what do we have to make it profitable, after which ideally putting the expertise inside these firms in a long run in order that these firms actually turn into AI firms and are in a position to transfer their very own initiatives ahead sooner or later.
How lengthy is the time period that they join usually?
Generally, 4 to 12 months.
It’s one thing we work out initially, relying on the complexity of the mission and what number of issues we’re making an attempt to resolve. We discover the longer, the higher. Machine studying tasks to do in 4 months might be difficult. There’s much more to it than simply constructing ML fashions. Heavily reliant on the info that is collected from the consumer that is handed over to us, that helps us construct the fashions. The longer we’ve, the higher it’s to iterate and cycle via the entire alternatives.
The work is experimental and exploratory in nature. Amii is a analysis institute; we will not at all times assure the end result. An extended runway simply provides us extra time to try this analysis and make it possible for we have exhausted our choices and pursued as many issues as potential as a result of it is exhausting for us to say, “This is the tactic that is going to work greatest.” You need to strive it and see.
What are some examples of difficult enterprise issues that your staff has labored on with these firms?
I alluded to it, positively knowledge preparedness is a giant problem. Ongoing {industry} notion of information preparedness is completely different than what a machine studying scientist would suppose is prepared for a machine studying mannequin. And entry. How simple is it for the consumer at hand over the info to us in a means that’s consumable for our ML fashions. That’s why we do like longer tasks as a result of it provides our staff time to work with our shoppers via these types of information preparedness challenges and set them up for fulfillment.
Garbage in is rubbish out, in the event you hand us rubbish knowledge, we will create a rubbish mannequin. We really want high quality knowledge. And there’s just a little little bit of a studying curve for shoppers. Industry notion, once more, of what high quality knowledge is, what are the examples that we have to see to have the ability to predict issues sooner or later. It’s only a literacy factor, ensuring that we’re talking the identical language, they perceive the restrictions primarily based off of no matter knowledge they’ve entry to after they perceive what is going on to set us up for fulfillment.
You want examples of what you are making an attempt to foretell in your dataset. If an occasion is absolutely uncommon, it should be exhausting for us to ever anticipate it taking place. We might construct a extremely correct mannequin of one thing that simply say 99% of the time correct as a result of it is by no means predicting the 1% time that one thing does happen. Again, simply ensuring that the consumer understands what we have to construct correct fashions.
We’ve seen even seemingly easy issues might be extremely advanced relying on their dataset. At the outset, having an preliminary discovery name with a consumer, we do need to anticipate the size of time that we are going to want. But typically after we begin peeling again the layers of the onion, we notice, no, that is way more advanced than we thought due to these knowledge complexities.
Other challenges, lack of dedication from subject material specialists wanted. When we companion with our {industry} companions, we actually want them to proceed to come back to the desk as a result of they’re the area specialists and normally the info specialists too. We’re not like a dev store the place we will simply take the info, construct the mannequin, and hand it over to them in the long run. It’s very, very collaborative. And the extra that our {industry} companions put in, the extra that they will get out as a result of they will have the ability to information us in the precise path, make it possible for the predictions that we’re making make sense to them from a enterprise perspective, that we’re focusing on the precise metrics, we perceive what success is for them.
We do want a multidisciplinary staff round us to assist the tasks and it takes greater than only one machine studying scientist to construct a profitable mannequin that is going to influence a enterprise positively. There’s numerous challenges. Those are those that got here to thoughts.
You personally consider that AI ought to be a pressure for good. What are some ways in which you suppose AI can positively change the longer term?
The factor I like most about my job is we work with shoppers from throughout all industries, fixing very completely different issues, however all of them are actually getting used for some kind of constructive change. And Amii has our principled AI framework that ensures that we’re doing simply that. From the contracting stage, we’re ensuring that the tasks that we’re engaged on with our {industry} companions are getting used for that constructive change in an moral means. All the tasks I get to see are getting used for good and positively altering the longer term.
One factor that involves thoughts, in Alberta most of the time now we’re coping with wildfire conditions in the summertime. This 12 months particularly, even in April, it was dangerous. We not too long ago partnered with Canada Wildfire. It’s a analysis group out of the University of Alberta. 40 years of climate knowledge tied to extreme wildfire occasions. Working with them to raised predict these occasions sooner or later so we will higher put together the sources that is likely to be wanted, have the groups go in and mood the environments earlier than it will get to a stage the place the wildfires are raging. I believe that is simply being in Edmonton, I do not know in the event you had been right here final week, nevertheless it was very smoky.
When I arrived Sunday evening (May 21, 2023) it was fairly smoky.
It’s devastating. It ruins communities. It takes folks’s houses away. Having to breathe particulate within the air is not nice, however the devastation may be very immense. That’s one attention-grabbing (mission) that is near all of our hearts.
Another space we’re working in is the agriculture house. How are we going to feed our rising inhabitants? We’re working with the National Research Council on a protein abundance downside. Trying to verify the crops that we’re rising have larger protein content material to feed our rising inhabitants and utilizing machine studying to have the ability to make these predictions.
Reducing emissions is one other extremely popular one. Working with firms within the oil and fuel sector to make it possible for the processes and techniques and instruments which can be used are as environment friendly as potential. We’re working with a water therapy plant out of Drayton Valley, which is a small city in Alberta, ensuring that that water therapy plant is working as effectively as potential and that we’re creating as a lot clear water for the neighborhood as potential. Precision medication as properly.
The checklist goes on. Literally, each firm we work on its these types of tasks, these types of causes. It’s exhausting for me to select a favourite as a result of when you concentrate on it, all of them have the likelihood to have a extremely constructive influence on the longer term.
What is your imaginative and prescient for the way forward for AI or robotics?
My publicity to robotics has actually been within the provide chain. It’s the place robotics are already getting used, nevertheless it’s additionally how can we improve them with AI to construct on present techniques and automation, once more, via extra environment friendly processes? The provide chain is clearly fascinated by rising throughput, fulfilling extra orders extra rapidly, and extra environment friendly decision-making. On the robotics aspect of issues, once more, my publicity has been constructing on prime of present robots to make them smarter and higher.
I believe extra typically, the longer term from what I see {industry} doing remains to be very human-centric. Robotics are used as a device, as an augmentation to people. Maybe robotics being deployed in circumstances which can be harmful to people the place we should not be uncovered to the environments. Robotics are a fantastic substitute for us in that case to maintain us safer. There’s additionally actually cool analysis being carried out by our fellows and bionic limbs, so simpler management and motion of people that do want that assist. All very a lot nonetheless tied to people and their use of those instruments however making it simpler for them to make use of and making their lives simpler via these new techniques.
In phrases of the way forward for AI typically, that is simply such an attention-grabbing time to be on this house. Industry is lastly getting it that AI is right here and it’ll change every thing and you may both lead or be led. I believe one in all Amii’s visions is to have each firm snug with the expertise, conscious of what it could possibly and can’t do, and actually prepared to experiment and iterate on implementing it of their enterprise to resolve a few of their hardest issues.
Up till now, I believe perhaps there was a notion that it was simply tech firms that had been AI and ML customers, however now it is changing into extra obvious that ML might be deployed in primarily each group. It’s not at all times the precise reply, however there’s normally a use case for it. I’m hopeful that the longer term is firms changing into pure AI firms themselves by getting extra literate and accustomed to the expertise and conscious of how they’ll use it for his or her enterprise.
Thank you for the superb interview, readers who want to be taught extra ought to go to the next sources:
