Mentoring the Workforce of the Future and Reverse-Mentoring the Workforce of the Present, with Steffani Silva


Steffani Silva (left) and Danny Nsouli (right). Graphic: Danny Nsoulii

Interviewer: Danny Nsouli

Welcome to the latest installment of the Knowledge-Driven Podcast. In this series, Cyber Security Software Engineer Danny Nsouli interviews technical leaders at MITRE who have made knowledge sharing and collaboration an integral part of their practice.

At MITRE, everyone is always learning. Education plays a big part in our technical innovations, so many of us choose to dedicate time to teaching and mentoring. In this installment of the MITRE Knowledge Driven Podcast, Data Analytics Engineer Steffani Silva, gives us an inside look at the inner workings of Generation AI Nexus, a collaboration among MITRE, universities, and government that educates students, as well as professors, on the power of artificial intelligence and accessible data.

A resource mentioned in this interview:

Click below to listen to podcast:


Podcast Transcript
Danny: 00:13 Hello, everyone. My name is Danny Nsouli, and welcome to MITRE’s Knowledge-Driven podcast. Today, I will be interviewing Steffani Silva, and we’ll be discussing her interest and involvement in mentoring through one of MITRE’s programs. Steffani, would you like to give some background about your experience and identify the program you’re involved with?
Steffani: 00:30 Sure. Sounds good. So, I am a data analytics engineer at MITRE. I’ve been around for about three years now, and I’ve done all kinds of work not related to education—which is what I’m going to talk about mostly today—[and] worked on healthcare projects and defense-related projects. But I do like to spend a little bit of my time anywhere that I can on just either mentoring or focusing a bit more on education. And I’ve done that at MITRE through a project called Generation AI, as well as some other places, like [the] MITRE Mentoring [program], and mentoring interns in the [MITRE] Summer Intern Learning Track, as well as outside of MITRE. So, I think today we’re going to talk a little bit more about Generation AI and some of the mentoring aspects.
Danny: 01:19 Great. And how did you initially get involved with Generation AI?
Steffani 01:23 I was actually taking a MITRE Institute class, so it really came full circle. I was sitting in as a learner, and I met Joe Garner who actually has a lot of formal training and experience as an educator and in the education space. So, he approached me, and we just started chatting. And after a while, he thought I might be a good fit for the program. So, I went and did sort of an internal interview with him and Marilyn Kupetz at MITRE [one of the Gen AI project managers], but I really was just intrigued by this idea of sort of creating literacy around AI. That’s something that I think I’d noticed a real need for working on other projects at MITRE. And so, it really sparked my interest in it, and it all just kind of unraveled from there.
Danny 02:05 And, mainly, who are the people that are being mentored in this case? Can you give a description on the target audience so to speak?
Steffani 02:14 Sure. So, with Generation AI, the goal is to teach students. So, we really focus on college students, a lot of times, undergraduate college students, especially in the non-STEM space. So, we might teach students in fashion design or marketing about AI, or at least having AI literacy. And we can talk more about what that means, but focusing on them, we’ve really expanded also to teaching graduate students, or even some high school students. And then a really interesting part of the work more recently has been this idea of training the trainer and thinking about the fact that we don’t just want to teach the students. We want to teach professors to be able to better teach all the students to come. And so, we really learned that professors were really afraid of teaching AI related concepts to their students, especially undergraduate professors not in a STEM space. And so, a lot of my job more recently was teaching professors to be able to teach their students-to-come in future semesters.
Danny 03:17 Do you find it more challenging teaching those that are not in the STEM space?
Steffani 03:22 Absolutely. I think, really the biggest thing, the biggest sort of hurdle there, isn’t that at all that those students don’t have a foundation to at least become AI-literate or aren’t able to; it’s that a lot of them see it as so far outside of their field, and they often have a lot of fear surrounding learning those topics. So, a lot of our job, I think, it’s just to find ways to (a) get people excited and interested in learning it, but (b) really just help them not be as afraid to approach things once they’re interested.
Steffani 03:58 So a lot of them start to think, “That’s so cool. I love this dashboard that’s using AI, and I’m so interested in how it’s working, but I would never understand that.” Right? There’s kind of that mentality, you’re seeing with a lot of, especially, freshmen students. And so, I think the big hurdle is to get them to have some confidence that when they come into one of these lessons or, as they’re coming into one of these lessons, they will be able to get what they want out of it. So, it’s sort of that confidence building is one of the big hurdles, I think.
Danny 04:32 And in your opinion, would you say that that hurdle is easier to get over with professors or with students?
Steffani 04:39 That’s a good question. Sometimes it is actually easier with students. I think part of it is that professors might be more ready to jump in and ask a lot of questions; they’re really engaged off the bat. But I think they also want a lot more information because they know they’re going to be this source of truth for their students. So, I think they feel that burden of needing to know things deeply, at a deeper level, than students feel they need to. And so, I think the hurdle is kind of a larger for that, right? They want a lot more information, which makes sense before they get up in front of a body of students who are going to ask them questions about that information. So, I’d say maybe a bigger hurdle with professors.
Danny 05:26 This is kind of a larger question, but can you take us through the process of how you generate these lesson plans and maybe comment on the differences depending on if it is a professor or a student that’s being taught?
Steffani 05:38 Sure. Sure. Well, generally, when we would create a lesson plan, I would be working with a professor, and things have changed. We sort of thought of ourselves as a startup, right? I mean, at the beginning, everything was extremely customized, tailored to each professor’s class. And then later we tried to get more… as we tried to scale, we became more generalized in how we built our lessons. But I think early on, it was about always starting this process by meeting with a professor and asking. From my end, it was just a lot of questions about what’s your background? But sort of more importantly, what are your students’ backgrounds? What are they comfortable with? Because, again, to cross that hurdle, I think the most important thing is to meet people where they are currently. So, my biggest thing I want to find out when I first meet with a professor or look at a class is where are these students?
Steffani 06:36 Where are they in their lesson material? What kind of foundation do they have? What kind of terms will they understand? And then I also want to see what the topics are that have gotten their students excited. So, I try to ask professors questions about where in their syllabus students tend to struggle or get excited about things? And then, we work together to find where a lesson can fit in, that would be most natural in their course. Because a big part of Gen AI is we don’t want to disrupt what professors are already doing. That whole purpose is to really meet them where they are. So really, it’s a load of questions on my end, and then, just working back and forth, it’s a very iterative process with professors. I like to rapidly prototype my lessons as much as I can.
Steffani 07:26 So, I would always be creating an outline, topics, working with them to find data sets. There’s a lot of data wrangling that people who are new to the data science world aren’t familiar with. So, they learn a lot, I think, just going with me through the process of creating a lesson and being able to iteratively meet with them, as I do that. I think they learn a lot about what goes into data science and all the sort of pre-work, the data wrangling, data munging, all of that. So, I work really iteratively with them until we start to come up with something that I can tell the professor’s excited about and that they think really meets their students’ needs. Then we create a lot of documentation about that so we can scale it to other professors as well.
Danny 08:18 And do you implement any testing elements in order to measure how well the information is being conveyed in these lesson plans?
Steffani 08:26 Absolutely. Yeah. Personally, I find it really important, not just in building lessons with professors, but especially anytime I teach, I find it really important to incorporate touch points as frequently as possible. So, when I’ve actually taught some of these lessons, I would try to use maybe, in virtual times, a QR code to be polling students with different questions related to the material. And they can just scan the QR codes off their screen from their phone and answer these poll questions. I’ve tried to incorporate those almost every 5 minutes into a lesson if I can, just as frequently as possible as a touch point. And at the end of every lesson, we always try to survey our students. In fact, we also would survey our professors. I’m a data scientist, so I really like metrics. So, I would always work with Joe Garner in the beginning, or we would work with different teams through Generation AI to survey students at a large scale and to have professors asking poll questions throughout their lesson.
Danny 09:30 That leads me to think that you focus a lot on interactivity to keep students engaged. Is that right?
Steffani 09:36 Yeah. Engaging is the key word. I think we all figured that out quickly. I mean, again, I don’t have the formal background in education, but I think it’s very apparent to anyone who teaches. But you want a lesson to be interactive because the more people are touching things hands-on… what I found is the more it builds competence in the people that you’re working with. So, whether it’s a student or professor, I try even for just our meetings to make sure that it’s interactive. So, with a professor, I might have them actually walk through… If I create a worksheet for students, I might actually have the professor walk through that worksheet and see if together we can answer those questions or make sure they’re competent doing the student piece before we move forward. So, anything I can do to just create more room for the person I’m teaching to do more of the talking, I think that helps solidify what they’re learning and also really starts to build a bit of confidence for them to keep learning.
Danny 10:46 And do you use any supplemental activities to, again, push for that interactivity to allow students to reinforce what they’ve learned?
Steffani 10:55 Yeah. Absolutely. So again, part of that interactivity is having, not necessarily just polls, but having something for the students to do and realize that they can really accomplish a big task in this space. So for instance, in the Summer of AI learning track, which is one of the learning tracks that our MITRE summer interns can choose, last year, I ran that with Ali Zaidi as part of Generation AI, and we had about 160 students come through, and we made sure everyone coming through was participating in teams, too. At the very minimum, even in a beginner level group, participate in a mini hackathon is what we called that.
Steffani 11:40 So, a 3-hour challenge. And then in our advanced groups, we try to give them even more room to run as a hands-on space. So, we give them about 7 hours just to really work throughout a couple of weeks, to really work through a hackathon challenge where they actually detect whether URLs are phishing URLs or not. And that’s something we’re actually working on right now, again, for this summer’s group of interns. And seeing where we can engage them even more and get them excited about solving a problem that aligns with MITRE’s mission and gets them to really address the problem head on, on their own.
Danny 12:19 And for the students, does Generation AI offer any online resources that they can refer back to that may also contain supplemental information to their lessons?
Steffani 12:31 Yeah. So, in Generation AI, and it’s actually a plug I wanted to throw out today, so thank you. I really love it. People check out our Generation AI Nexus [portal], which is that at And that’s where we maintain all of our lessons, all of the surrounding material for our lessons. So, when I mentioned [that] we create documentation for professors, really, we learned that scale. If we weren’t going to be working with professors on the ground, people who take our lessons, want a lot of supporting materials just as the instructor. And so, we worked with Joe Garner in the early days of Generation AI to create this sort of structure for training the trainer in terms of support materials for professors.
Steffani 13:18 So, you can see all of that at, and that’s sort of the final format we try to get our lessons into. I’d say it’s really different depending on where I’m teaching. Sometimes I’m teaching senior epidemiologists, sometimes I’m teaching a One MITRE Mentoring course, then sometimes I’m working with interns. And so, for each group that I’m working with, we try to find the best way to hand off our finished product. So sometimes, it’s a Git repo. Sometimes it’s just a Slack channel and code being shared.
Danny 13:53 Jumping off that note, since we’re talking about online learning, have the effects of the pandemic altered your teaching process at all with all the changes of remote learning?
Steffani 14:04 Well, it’s funny you ask. I actually hadn’t started teaching until right around the time that we went virtual because of COVID. So, I don’t know that, as far as me actually teaching people, I think I learned during the pandemic. But as far as creating lesson content, I would say that’s changed a lot because I was creating lesson content earlier on for professors. And so, we did have this whole moment of learning how to shift from being in person to being online and what that could mean. And if our lessons were going to be hybrid lessons, or if we would create lessons specifically for the online space versus a physical in-person classroom. So, it definitely did affect how we created our lessons, and we had some lessons ourselves, actually from outside sources, to come in and teach us about how we could shift our lesson content to be better suited for online classrooms.
Danny 15:09 Since we’ve talked about interactivity being a big part of your lesson plans, do you also apply a lot of collaboration elements to train students for the teamwork that comes with joining the workforce?
Steffani 15:22 Absolutely. Well, I always try to have a portion of the lesson, and in fact, a large portion of, say, 25% of the time or more, the students working together in groups to solve a problem using some of the skills they learned earlier in the lesson. So, if I can, I try to follow around maybe a 2-hour lesson, I might follow a format where for 20 minutes where we’re learning a concept together. And then for maybe 5-10 minutes, we’re sort of testing together all as a whole class that we understand that material. And then we might take a half hour to actually say, “Okay. Now you’re in groups and you’re going to do a very similar thing, but together in your groups, and we’re just going to come around and see if you have questions.”
Steffani 16:15 I mean, I see this working really well with students, but the funny thing is I actually see it working… I see it being even more effective with adult learners. So, some of the sponsors that I’ve worked with where we come in and try to teach in the workplace. I think this is actually even more effective for them because they realize already exactly what you said, that they’re going to be collaborating in the workforce. And so, they sort of start to use these people, I think, as references after the class as well, which is what we hope for. So, we always try to push that: Afterwards, if you’re working on something two weeks later and you need to apply this, please do reach back out to the people that you were working in groups with, with the other people who came today. And we’ve actually seen that start to happen. That’s really exciting.
Danny 17:03 And would you say that allows for more organic networking between mentees?
Steffani 17:08 Yeah. Actually, there is a lot of networking. I’d also say within the mentoring groups, that’s probably one of the biggest places I see that as well. So we assign little homework assignments in our MITRE mentoring sessions, and we always heavily suggest that people go off and work together on them. So I do think some of our mentees have started to reach out to each other and spend a little time together working on these mini assignments before they come back to us in two weeks. And I think it is always helpful to learn something with someone else, in part, because you might catch different things, you share your different perspectives and ways of doing things. And I think that also just makes people more excited about learning the content.
Danny 17:55 So, in terms of the long-term, do you see any changes coming to Generation AI with the goal of improving what’s being done already?
Steffani 18:03 One big change is that we’ve really started to scale. We’ve scaled a lot since I came on board the project over a year ago. And so, what that means in part is no longer just meeting with professors and customizing and tailoring everything to a professor that I’m getting to meet with day after day. That’s really fun, but it is much harder to scale, than being able to create sort of modular lessons, which is what we’ve been moving towards. So I think one big change is that the lessons are becoming more and more modular so that you can take bits and pieces of them. And that we can focus more in the future on this train the trainer program, where we can have professors go through some sort of training themselves to be able to understand how to pick sort of pieces of this lesson up that fit to their classroom best.
Steffani 19:02 So, another thing that’s going to change in the future is that we’ve created this consortium. So, a network of professors and partners that Generation AI is working with who are really becoming more and more self-sufficient. So that’s been awesome to watch. We’ve been working with partners like Purdue and UTSA [University of Texas at San Antonio] among many others. For them to really be able to keep this going themselves, I think we’re definitely trying to work ourselves out of the picture there, right? And create this fully self-sufficient Consortium so that not only do students build networks as they’re working on a lesson, but also professors can network with each other across schools. And we’ve seen time and time again, as professors start up programs, or just start teaching Generation AI lessons in their schools, that a big part of the help for them is to have other professors to reach out to. And so, I think the consortium really does that. So that’s another growing change, I think, as Generation AI progresses.
Danny 20:06 So, to wrap up, based on your mentoring experience, do you have any general advice for any of our listeners that may be educators as well?
Steffani 20:15 I’d absolutely say, work in as many touch points and as much interactivity as you can, and really let people get hands-on, whatever you’re doing. So if you want to teach people about data, I like to start by having people touch data, create data, ingest data. I think the minute they realize they can do it, they get more interested. I think another thing is to make as few assumptions as possible about the learners. I think that people often are afraid, or I was when I started teaching, afraid to sort of insult the learners by explaining maybe what a box plot is at the beginning of my lesson. Because a lot of people might look around and think, “Please, I know what that is and now I feel insulted.” But honestly, I’ve learned that even if there’s one person in the room that you can capture by just reviewing something, I think it’s actually been much more appreciated than anything.
Steffani 20:17 So, I heavily recommended making as few assumptions as possible and reviewing, if you think there’s even a percent chance that someone might not recall some foundational information, cover it and you’ll pick people up along the way by doing that. And then I really also encourage just reinforcing what you’re teaching people, by telling people what you’re going to tell them, telling them what it is. Once you’ve told them that you’d explained it, so we check things off a list as we learn in a lot of my classes and say, “Okay, I’ve learned that.” And we pause. And then, again, reviewing at the end what we learned and really allowing for that question period, and all that engagement. With that, I might also throw out that if anyone is interested in collaborating, or if anyone wants to take a look at the lessons in the Gen AI Nexus [portal], then please feel free to look over them and reach out to me, reach out to someone in our Gen AI team. I love talking to other people who had experienced creating content or teaching. And I’d also love to hear feedback on some of the lessons you’ve got.
Danny 22:33 Sounds good. Well, Steffani, it’s been a pleasure hearing about your experience and mentorship. I’d like to also give a quick thank you to MITRE and the Knowledge Driven Enterprise for making this show possible. And again, thank you, Steffani, for taking the time to discuss your work with us.
Steffani 22:47 Thanks, Danny.


Danny Nsouli is an Associate Cyber Security Software Engineer. He has a passion for computer graphics and enjoys learning about front-end solutions for consumer-facing project components such as data visualizations.

© 2021 The MITRE Corporation. All rights reserved. Approved for public release.  Distribution unlimited. Case number 21-1431

MITRE’s mission-driven team is dedicated to solving problems for a safer world. Learn more about MITRE.

See also:

So You Want to Think Like a Data Scientist? The Importance of Visualizations in the Data Science Workflow

Enhance Your Data Science Toolkit: Add-Ons and Updates to the Most Commonly Used Tools

Interview with Ali Zaidi on Designing Lessons in Artificial Intelligence

Humble Beginnings

Coming Back to Make a Difference, Find a Passion, and Change the World

Mentoring the Workforce of the Future: The Emerging Technologies Summer Student Program

When AI and Psychology Meet, Insights Emerge

Creating an AI-Savvy Workforce for a Strong Future

MITRE’s Pearls of Expertise at FIU ShellHacks

Interview with Dr. Michael Balazs on Generation AI Nexus

The World as It Will Be: Workforce Development Within and Beyond MITRE

Jen Choi and Josh LeFevre and the power of “Yes, And”

Getting Students Excited About STEM (and MITRE), with Willie Hill

Theodore Wilson: Thinking Like a Turtle

Interview with Jackie Morin on her journey from intern to senior engineer

Interview with Jay Crossler on why passion is the key to success

Interview with Dr. Philip Barry on blending AI and education

Interview with Dan Ward, Rachel Gregorio, and Jessica Yu on MITRE’s Innovation Toolkit

May 22, 2021


Pin It on Pinterest

Share This