By Francie Fink
Dan Harmon is the Director of Data and Systems with the University of Illinois Sponsored Programs Administration (SPA). SPA is a service unit under the Office of the Vice Chancellor for Research and Innovation that supports externally sponsored research projects throughout their lifecycle. Dan oversees a team that focuses on data analysis, training and outreach, cash management, sponsored billings, and software development. Recently, Dan has been investigating the use of AI in sponsored programs administration. We had a conversation last month, where I had the chance to ask Dan about the data science dimensions of his work in research administration, and the broader implications of AI-driven innovation in this field.
Key takeaways:
- SPA’s Job– The office handles proposals from submission through contract negotiation, billing, subrecipient payments, and compliance oversight.
- A Data-driven SPA– The team provides metrics to administrators, runs scenario analyses on potential funding cuts, and compares historical data to spot trends in proposal volume, success rates, and spending.
- An AI-powered SPA– Tools include AI checks to ensure proposals align with lengthy sponsor guidelines, an AI-powered chatbot for policy questions, and automated compliance monitoring to track where funds are going.
- New Tools – Streamlining of budget creation will shorten submission timelines and advancing automated invoicing systems will reduce errors and improve tracking of payment from funding resources.
What roles does SPA play in supporting research at Illinois?
We’re the research administration office here on campus, and we manage the grants and contracts awarded from external sponsors. That could be the federal government, or the state of Illinois, private companies, foundations, even foreign companies and governments. Our office handles the full life cycle of those awards. We submit research proposals to sponsors, negotiate the terms of the awards, and once they’re executed, we manage them. That includes billing, working with subrecipients—partners who collaborate on research—and making sure we pay them in a timely manner. We also have an audit and compliance arm that ensures everything is done correctly.
I oversee the Data and Systems team. We have a bit of a non-traditional setup because not everyone on my team is technical. For example, we have an accountant who handles financial reporting and accounting. There’s also someone who manages invoicing—making sure bills are generated and paid on time. There’s a cash management team that processes incoming payments. Believe it or not, we still get checks in the mail! We also do training for the entire university. We run a course called SPaRC’Ed (Sponsored Programs and Research Compliance), which is open to research administrators across campus. It walks them through how to do research administration correctly, according to the various legal and accounting guidelines we have to follow. There’s a lot to stay on top of!
On the technical side of things, my team is involved in data analytics, software development, and database management. For the past couple of years, a major focus for our team has been AI—figuring out how to make our office more efficient with less resources. AI can be a really powerful tool for that. We see ourselves as national leaders when it comes to using AI to improve operational efficiency in our field. I’m very lucky to have an excellent team with lots of years of experience. The chat bot for our office is even named after one of my employees!
I’d love to hear a little more about your personal background—how you got into this and what your path here looked like.
I actually come from a computer science background. After I graduated, I didn’t go into Computer Science right away—I went into accounting for a year. Then in 2005, I started working at the University of Illinois, first with the Illini Union as a software developer. And then around 2009, I moved to Enrollment Management. There, I worked on all the admissions-related systems. I developed the application that students use to apply to the university, plus other internal systems for reviewing applications and managing things once students are admitted. And then about three years ago, I started here in SPA as Director of Data and Systems.
When you talk about research administration, who are you typically working with? Is it mostly faculty with research projects, or more with university administrators who help secure funding?
It really runs the gamut. A lot depends on the department and how much support they have on their own research administration team. But a typical life cycle starts when a faculty member has an idea and they start looking for funding. They’ll find a request for proposals from an agency or institution, and then they’ll start putting together a proposal. At that point, they’ll usually send it to their department’s research administration office to fine-tune the budget, which can be particularly tricky, and they’ll make sure everything is in order.
Once that’s done, it comes to us. Our office does a full review to make sure all the I’s are dotted, T’s are crossed. Then we submit the proposal on behalf of the faculty member via the sponsor’s system. So really, depending on the internal support within a department, that interaction might come directly from the faculty member, or it might be routed through their department’s office. It just depends on the resources available in that particular unit.
Can you tell me more about your data analytics team? What kinds of projects are they working on?
A couple of big things, really. One of our goals is giving administrators the tools they need to understand how they’re doing—metrics that show performance and where there’s room for improvement.
And then there are bigger-picture analytics we do, too. Agencies and granters can cut funding, and we can run scenario analyses: what happens to the university’s funding if proposed cuts go through? So we can see if our F&A rate in a given area goes down 15%, what is the impact on us financially and how can we plan for that?
You mentioned you’ve been working a lot with generative AI lately. What’s the conversation about its use in the research administration community more broadly?
I’ve always been interested in AI, even before generative AI became mainstream. Before this, when I was working in admissions, we were looking at how to predict the performance of a student based on their essays and other work samples. That’s hard to do with traditional data analytic methods, so we worked with machine learning even back then. When I moved into this role, we had a similar challenge, but with staff workloads—we needed to know whether we were balancing our workload efficiently. We wanted to make sure each staff member had a fair portfolio of work that didn’t overload them. That also lent itself well to machine learning models.
Then, when generative AI hit the scene, it coincided with a national conference. My boss, who sat on the board of that conference, had the idea to host an executive leadership meeting before the conference, to get leaders from universities in the room to start talking about its use in our field of work. That idea snowballed into making generative AI the focus of one of the keynote addresses for the entire event—and I was volunteered to deliver that talk. That kind of launched me into this role where I’ve been giving presentations, leading webinars, and speaking at conferences about what generative AI can do in research administration.
What I’ve seen in the last three years is a pretty dramatic change. Early on, if you’d ask a room, how many people are using generative AI? Maybe one or two hands would go up. Now, probably 50% of people would raise their hand, and I’d guess in the next month it’d probably be 75%.
Where do you see AI being the most useful for research administration?
You can really use it across the entire life cycle. Some of the early use cases have been with contracting and negotiation. For example, when we get a contract, we can use AI to scan through it and flag things we’d typically redline—terms or conditions we would want to change. Those contracts can be hundreds of pages long, so this can be a huge time saver.
With proposals, we can use AI to compare what a faculty member submits with the sponsor guidelines to ensure they’re aligned. Those guidelines can be 300 pages or more, so again, it saves a ton of time.
We can also use it when the award comes in. There’s a lot of complexity to the workload that comes with an award. Instead of having someone manually enter all of the information into our system, AI can read the documents, pull out the relevant data, and populate the system automatically.
I briefly mentioned our chatbot already, but that’s another cool thing we’ve been working on. The AI-powered chatbot we have is trained on our policies, procedures, and knowledge. It’s tailored to the university, so anyone can ask it questions any time of day, and it can answer.
So beyond collecting data on secured funding, what other kinds of data do you regularly analyze to help guide faculty who are seeking awards?
We look at a lot of historical data. We can compare, for instance, where a department is this year versus the same point last year. If spending is down, we’ll dig into whether they’re submitting fewer proposals, or if the success rate—the “hit rate”—has dropped. Then, we can have a conversation about where to intervene, in order to make those proposals more robust.
There are also some interesting analytics on the compliance side of things. For instance, if you’re 80% of the way done with a project but have spent less than 20% of your budget, there’s maybe a possibility that you have funding from another source that you haven’t disclosed. We can show in a heat map chart which awardees are the ones that we need to take a closer look at. To be clear, it usually ends up being that there’s a good reason for the discrepancy, but there’s always the potential for noncompliance.
Can you look at what topics or kinds of proposals tend to get funded so faculty can tailor their applications better?
We haven’t been able to do this extensively yet. That’s because we’d need to dig into the full award documents to get meaningful insights—extracted from PDFs and different sources and analyzed. We do look at the title of awards, but they are often vague and don’t tell us the whole story. That’s where generative AI could be a huge help—it can summarize the core science and extract keywords or themes so we can run trend analyses and identify “hot topics.” It makes a process that is usually time-intensive and expensive easier.
So who is actually supporting departments when it comes to data management—like tracking proposals and helping with awards?
A lot of departments have their own research administration group that supports faculty. Some colleges have a shared service model where they will have a larger group that supports the whole college. Those administrators come to our office for training—they’re our main audience for the SPaRC’Ed program. We also host a big annual conference with about 300 attendees, plus quarterly meetings that bring in 300-400 people. The whole research enterprise on our campus is huge.
You said you oversee some software development. What sorts of high-level projects are you working on there?
We have an internal tool that manages all new awards, creating work queues and managing workflows. Our sub-recipient monitoring portal is used to keep track of sub-recipients on awards. We are getting ready to release a new invoicing portal within that section of the tool, so that sub-recipients can log in, submit invoices, and get paid. We also work on a lot of automated notifications, based on our datasets. This year, too, we’re working on developing a budgeting tool that will allow faculty and researchers to create budgets more efficiently, rather than using their own spreadsheets. It will be integrated with AI. Hopefully, they’ll be able to spend less time working on budgets, which typically account for a large portion of the submission timeline.
Looking ahead, what are some projects that you’re excited about working on?
Well, our improved budget tool will be a game-changer, both for us internally and faculty. Right now, faculty submit budgets in all sorts of spreadsheet formats, which are time-consuming for our proposal team to reformat. Because of that, we don’t really have good access to granular, line-item budget data. We can follow the budget through the entire lifecycle of the project and analyze budget trends across departments. We can monitor compliance, like verifying salary rates and prefabricated equipment spending. Or, we might want to automate multi-year salary adjustments, which we will be able to do easily.
We’re also going to be working on an invoicing system for our sponsors. Right now, invoicing with our sponsors happens manually, and we have someone on staff who corrects errors. With the new system, we can control errors on the front end, which will save us a lot of time on the backend. It will also allow us to track key performance metrics like payment timelines, accounts receivable, and dedicated staff time towards each award. That way, we can identify bottlenecks, improve processes, and make things more efficient.
Do you see value in having people throughout the research lifecycle be comfortable with data, especially as it relates to decision-making?
Yes, and I think that the way we visualize data is very important for that. Strong visuals reduce the reliance on extensive data literacy. When reports are designed well, for instance, trends and outliers become easy to spot. We use colors, layout, and the sizing of information very intentionally. We build tailored Power BI dashboards for different audiences—executive leadership versus departmental staff, for example. And because proposal information is very confidential, we have secure systems and protocols in place to keep researchers’ ideas and projects protected.
Learn More and Get Involved
Learn more about Sponsored Programs Administration here. To explore SPaRC’Ed’s courses and materials, visit the dedicated webpage.
Contact the Office of Data Science Research if you’re aware of other people or resources we could profile here. ODSR is a campuswide convening organization that facilitates collaborations, resource sharing, and public engagement focused on data science research activities at the University of Illinois.