In this edition of Cosmographer Corner, we are pleased to highlight Tuja Khaund, a former Graduate Research Assistant at the COSMOS Research Center. She earned her Ph.D. in Information Science from the University of Arkansas at Little Rock. During her time at COSMOS, Tuja made significant contributions to research on online influence operations, social bot detection, and coordination analysis in information campaigns. Her work involved developing data pipelines to analyze social media behavior using graph theory, network science, and natural language processing, as well as contributing to DARPA-funded projects focused on identifying bot coordination and influence flows in online networks.
Today, Tuja serves as a Senior Data Scientist at Walmart Global Tech, where she leads projects across natural language processing, graph analytics, computer vision, and machine learning deployment. Her professional success reflects the strong foundation in experimental design, analytical rigor, and interdisciplinary collaboration she developed during her time at COSMOS.
“I am proud of the accomplishments and the many contributions Tuja has made to the COSMOS Research Center. We celebrate her academic journey and wish her continued success at Walmart Global Tech,” said Prof. Nitin Agarwal.
We recently connected with Tuja to learn more about her current role, her experiences at COSMOS, and her advice for future Cosmographers.
Could you tell us a bit about your time at university?
I first came to UA Little Rock in 2013 as an exchange student. I transferred shortly after and completed my B.S. in Computer Science in 2015. I moved straight into the M.S. program in Information Science and graduated in 2017. During my master’s, in 2016, I began research that ultimately led me to pursue a Ph.D. under Prof. Nitin Agarwal. I completed my Ph.D. in Computer and Information Science in 2021. Looking back, COSMOS provided a clear progression, from coursework to independent scholarship to research, and the mentorship I received from Prof. Agarwal set the foundation for the work I do today.
How did COSMOS fit into your university education career? How did you come across COSMOS, and what were you studying when you joined COSMOS?
In my first semester, 2013, a senior introduced me to Prof. Nitin Agarwal. I later took Prof. Agarwal’s courses on Web Programming, Social Computing, and Information system Analysis, stayed in touch, and in 2016, I joined his research team, COSMOS. COSMOS quickly became the anchor of my graduate training. I worked on DARPA-funded research focused on social bot detection and understanding bot behaviors, which taught me how to design studies, build datasets, and collaborate across disciplines. That project naturally evolved into my Ph.D. work, where I studied social bot coordination in online information campaigns. In short, Prof. Agarwal and the COSMOS served as the bridge between coursework and real-world research, providing me with the mentorship, rigor, and community that shaped the rest of my academic path.
How did COSMOS contribute to your career and program at UALR? What was Prof. Agarwal’s role in your journeys during and after?
I joined COSMOS in 2016 during my second year of the M.S. After an early experience in a different research group where I didn’t quite find my fit, I was unsure how to shape my research. I met with Prof. Nitin Agarwal to discuss my interests, especially data science, which at the time didn’t have a dedicated program at UALR. We revisited Prof. Agarwal’s Social Media Mining & Analysis course I’d taken in 2015, where I first saw how graph theory and network science could be applied to social and information networks. Prof. Agarwal then introduced me to one of his Department of Defense funded projects, particularly the project on social bots funded through DARPA. The problem space clicked for me immediately as it combined real-world social media behavior, questions of misuse and manipulation, and the kind of analytical “puzzle-solving” I enjoy. Prof. Agarwal and COSMOS gave me a clear research home, strong mentorship, and the methodological grounding that carried me through my master’s, into my Ph.D., and ultimately into a data science career.
How would you describe the “research pipeline” that you worked on while at COSMOS? In other words, what was the specific area in which you researched?
My work focused on online influence operations, with two complementary threads: detecting social bots and mapping how they coordinate within information campaigns using principles from graph theory and network science. I built pipelines to ingest public social media data, clean and enrich it, and model behavior using a combination of temporal, content, and network features. We combined high-precision heuristics with supervised models for social bot detection and then constructed multi-layer interaction graphs, such as retweet/reply/mention networks plus hashtag and URL co-sharing, to surface communities and influence flows. A major emphasis was on coordination detection: synchronized posting windows, near-duplicate content bursts, and shared payloads across accounts. Alongside Python/NetworkX/Gephi, I used Maltego for social cyber-forensics, including a study that uncovered coordinated “blog farms” during NATO’s Trident Juncture exercise. I also leverage NLP/NLU to analyze Telegram discourse on Ukraine’s political climate and sentiment toward politicians. The primary outputs were peer-reviewed publications and sponsor briefings, supported by analyst-ready visuals.
How did your time at COSMOS shape your career path or personal growth?
UA Little Rock shaped how I think and work. COSMOS trained me to turn open-ended questions into measurable studies, build reproducible pipelines, and communicate results clearly. Publishing and sponsor reviews taught me to defend choices with evidence and accept feedback quickly. Mentorship from Prof. Nitin Agarwal and the diversity of the team gave me leadership roles and a practical sense of impact. Those habits of rigor, collaboration, and clear writing carry directly into my data science career, guiding what I prioritize: useful methods, transparent results, and work that matters.
Since leaving COSMOS, what roles/positions/jobs have you had? What is your current work? What positions did COSMOS and your classes at UALR best prepare you for?
Since leaving COSMOS, I have worked at Walmart Global Tech as a Senior Data Scientist, leading projects across NLP, graph analytics, computer vision-based signal fusion, and production ML, spanning from data engineering and modeling to deployment and stakeholder communication. Alongside those efforts, I began developing retrieval-augmented generation (RAG) components such as document pipelines, retrieval and evaluation frameworks, which matured into a broader focus on agentic workflows and citation-grounded question answering. COSMOS and my coursework at UA Little Rock prepared me especially well for roles that combine research with delivery such as Data Scientist, Applied Scientist, and Research Scientist. The lab’s emphasis on experimental design, network science, and clear communication translated directly to designing end-to-end systems, explaining trade-offs to non-technical partners, and shipping models that solve real problems.
If you had to describe the most momentous event at COSMOS, what would it be? (Such as a specific conference you presented at, or a specific paper presented, or a project you worked on, and your favorite memories or experiences as a student at COSMOS).
The most momentous event was presenting my first paper at the SBP conference in Washington, D.C. It was my introduction to the rhythm of a research conference such as poster sessions, Q&A, and the kind of candid hallway feedback that sharpens ideas. I also met many of our sponsors and leaders from industry and government, which helped me see the “big picture” as to how our methods translate into real-world needs, priorities, and constraints. Those conversations shaped my next set of research questions and gave me a clearer sense of impact. I met peers who later became collaborators on our MURI work, and as my first time in D.C., that trip doubled as real team bonding. We decompressed after sessions, compared notes, and got to know one another outside the lab. That experience gave me confidence as a researcher and cemented relationships that have lasted years.
What advice would you have for current Cosmographers? (Such as looking for conferences to present at, jobs to apply for, and recommending fields that social computing research/skills can be applied to.)
My advice to my fellow cosmographers is to aim for workshop-to-conference momentum at venues such as SBP-BRiMS, ICWSM, WebSci, SocInfo, and ASONAM. Prof. Agarwal maintains a strong vision and a roadmap that will help you target the right outlets. When possible, extend conference work into a journal, once the core idea is validated. Make your work easy to evaluate by including a clear methods section, a documented data schema, strong baselines, and results that feature clear visualizations. Be candid about gaps and next steps; thoughtful limitations strengthen a paper. For the job search, maintain a focused resume that aligns your skills with roles in data science, applied science, AI/ML engineering, and trust and safety. Network with intention, ask leaders about their career paths, read beyond your domain, and consistently showcase your work. Stay close to emerging research and industry trends by scanning recent proceedings and preprints, tracking major releases, and turning one new idea each quarter into a shareable project.
Are you still connected with classmates, faculty, or the alumni network?
Yes. I keep up with classmates, stay connected with faculty (including Prof. Nitin Agarwal), and engage with the alumni community for collaboration, mentorship, and opportunities.
What’s next for you, any exciting goals or projects you’re working on?
I am focused on advancing retrieval-augmented generation and agentic workflows that produce reliable, citation-grounded answers. My near-term goals are to improve retrieval quality with graph-aware signals, strengthen evaluation with stress tests and ablations, and package the work into reusable components. In parallel, I am exploring speech and audio tasks, including robust voice activity detection (VAD) and practical ASR pipelines for noisy environments. My goal is straightforward: to develop useful systems that connect research methods to real-world outcomes.