SPARC 2026: Inside Penn Engineering’s Summer Tech Project Competition for Online Students

A Showcase of Innovation at SPARC 2026

Each summer, graduate students in Penn Engineering’s online MAS-CS and MCIT programs have an opportunity to take what they have learned beyond the classroom and build something of their own. Through the SPARC (Side Projects for Advancement, Refinement, and Collaboration) competition, students form small teams, develop an original technical project, and spend 12 weeks turning an idea into a working product.

Launched in 2021, SPARC gives students space to practice existing technical skills, experiment with new tools, and gain experience building a side project from the ground up. Teams work alongside Penn Engineering alumni mentors as they define their ideas, divide responsibilities, solve technical problems, and refine their projects throughout the summer.

The MAS-CS degree is designed as a computer science pathway for learners from a wide range of academic and professional backgrounds with minimal prior computer science experience. SPARC offers an important opportunity for our online graduate students to supplement the skills and knowledge they develop through the rigorous MAS-CS curriculum with practical, hands-on experience putting new skills to the test in real-world environments.

The 2026 SPARC projects submitted by competing teams reflected a wide range of interests and real-world applications. Students developed tools related to healthcare, career development, travel, education, language learning, community support, home renovation, AI reliability, and more. Across the projects, teams explored how technologies – including generative AI, machine learning, mobile development, data analysis, and full-stack engineering – could be applied to practical challenges across a wide range of fields.

The projects below highlight what this year’s teams built and what students learned along the way.

Meet the 2026 SPARC Teams

Winning Team

Chalice: Interactive, AI-Powered Medication Management Training

Members: Anna Liao, Nanda Putra, Jodie Tai, and Joshua Weiss
Mentor: Qian Wang (MCIT Online ‘25 Alum)

Medication reconciliation requires clinicians to review a patient’s medications and consider how each drug, dose, and combination fits the patient’s health needs. The Chalice team created an interactive training environment where healthcare students can practice making those decisions with simulated patients. As students adjust a medication regimen, the system responds with a project-defined Net Therapeutic Benefit score that reflects factors such as appropriate treatment, kidney and liver function, drug interactions, and patient-specific metabolism.

Chalice also gives healthcare students detailed feedback on the reasoning behind their results. Its scoring engine records the benefits, risks, interactions, and contraindications associated with each change, creating a consistent foundation for the AI-generated educational feedback students receive. The team also built clinically plausible synthetic patient profiles so learners encounter cases shaped by interconnected diagnoses, demographics, organ function, and medication regimens.

For Jodie Tai, developing the application reinforced the importance of carefully evaluating AI within a working system. “Working with an LLM as part of an actual product really highlighted how unreliable AI output can be; like any external service, it needed to be checked and handled carefully.” Her work across the frontend also pushed her to consider the complete student experience, from building a medication regimen to reviewing results and past sessions.

For Anna Liao, SPARC offered a chance to put her coursework into practice with a team. “While my coursework has given me a foundation in computer programming, data structures, and software design, I wanted to gain hands-on experience building a larger project collaboratively and see how these concepts come together in practice.”

JourneyMind: Capturing the Story Behind a Child’s Growth

Members: Felix Deng, Crystal Ma, Yvonne Qiu, and Jade Wang
Mentor: Andrew Kuck (MCIT Online ‘25 Alum)

Childhood memories are often preserved through photos, but JourneyMind focuses on something less visible: the ideas, questions, and behaviors that reveal how a child is changing over time. The team developed a mobile app where parents can quickly record moments using written entries, dictated notes, or images. JourneyMind then helps organize those observations into recurring themes, while also giving families the option to compile favorite memories into a downloadable storybook.

The team also considered how to use AI responsibly when interpreting something as personal as a child’s development. Instead of asking AI to write open-ended assessments, they designed a more controlled process in which the technology sorts observations into predefined categories and the app calculates the summaries parents ultimately see. Users can also return to the original journal entry behind an observation, making the connection between an insight and the moment that inspired it clear.

Reflecting on the experience of creating a product for real users, Jade Wang shared, “My biggest takeaway is that technical capability gets a product built, but thinking from the user’s side is what makes it something people will keep using.”

For Felix Deng, the project also demonstrated how different collaborative development can be from working independently. “What stayed with me in the end was less the code than the habit of working transparently — communicating early, leaving things clear enough for the next person, and trusting my teammates with the parts I wasn’t holding.”

Reddit Virality Predictor: Guidance for Posting on Reddit

Members: Sarah Gillis, Kristin Lai, Zoe Tian
Mentor: Jinyeong Park (MCIT Online ‘25 Alum)

Posting successfully on Reddit involves choosing the right community, timing, and approach for a particular audience. The Reddit Virality Predictor team developed a web-based tool that helps creators, marketers, and small businesses make those decisions using historical engagement data from more than 96,000 posts across 48 subreddits.

Because online engagement is difficult to predict, the team designed the platform to show how much evidence supports each recommendation. Results are assigned confidence levels, helping users understand when the underlying data is stronger or more limited. The system considers factors including subreddit, posting time, content format, sentiment, emotion, and readability, while an AI layer helps interpret users’ topics and present the findings in accessible language.

Kristin Lai helped build the project’s data warehouse before moving to the frontend. “Transitioning to the frontend, I took ownership of the Streamlit application build to bridge backend machine learning outputs with an intuitive user experience.”

For Sarah Gillis, the project revealed how a non-technical background could contribute to engineering work. “My training in topics like research methodology and linguistics became an asset in this project.”

Speakeasy: Building Confidence in Speaking English Through Practicing Real-Time Conversation

Members: Hon Wai Chan, Xiaojian Chen, Renee Guo, and Lulin He
Mentor: Xintian (Emily) Yu (MCIT Online ‘22 Alum)

Speakeasy focuses on a familiar challenge for many English learners: knowing the language on paper and struggling to use it comfortably in the moment. The team created a mobile app where users practice short spoken conversations with an AI partner in situations such as networking events, job interviews, first dates, and everyday social settings. Each session gives learners a chance to rehearse the kinds of interactions they find most difficult and return to the same scenario for another attempt.

After a conversation, Speakeasy reviews the transcript and gives users feedback on areas including fluency, confidence, tone, conversational stamina, and cultural fit. Learners also receive suggestions for stronger phrasing, follow-up questions, and ways to keep a conversation moving. The team designed the experience around repeated practice, giving users a clear way to apply feedback and try the same interaction again.

Building the app pushed the team to think about conversation from both a technical and human perspective. Lulin explored what makes an AI exchange feel natural, including response timing, tone, and the ability to introduce new topics. Reflecting on the experience, Lulin shared, “This experience helped me better understand both the technical and user-experience challenges of building an AI-powered application.”

For Hon Wai, working on Speakeasy also changed how he thought about collaboration and learning. “I found that explaining these ideas to others pushed me to understand them more deeply myself, and it strengthened my own foundation in the process.” Renee reflected on the challenge of moving from an idea to something people could actually use: “Working on SpeakEasy made me realize how challenging it is to turn an idea into a real product.”

OurAtlas: Turning Saved Places Into Travel Plans

Members: Jay Deng, Qihang Fan, Zijin Wang, and Hongjin Yang
Mentor: Wenhao Guo (MCIT Online ’20 Alum)

Travel inspiration often ends up scattered across screenshots, social media posts, messages, and notes. The OurAtlas team created a mobile map-based app that brings those discoveries together in one place. Users can import content from links, screenshots, videos, or written notes, review the places identified by the app, save favorites, and organize them into trip itineraries.

The team also built Atlas AI, which works with a user’s own collection of saved places to answer questions and help organize a trip. A Discover feature surfaces nearby events, while the itinerary builder lets travelers arrange destinations across different days and times. Users review extracted places and approve proposed changes before new information is added to their collection.

Zijin Wang credited early conversations with the team’s mentor with helping them navigate unfamiliar decisions. “Whenever we hit a question or a decision we weren’t confident about, we brought it to our mentor before it turned into a blocker, which has kept us from spending days going down the wrong path.”

For Qihang Fan, working across different areas of the project offered a chance to learn from his teammates. “It was an amazing experience to communicate when different members have different skills.”

Village: Coordinating Community Support During Major Life Events

Members: Ila Keshishian, Brian Law, Rachel Wu, and Amy Yang
Mentor: Esteban Morales Oliveros (MCIT Online ‘25 Alum)

Major life events such as illness, bereavement, or the arrival of a new child often bring an outpouring of support from friends, family, and community members. Village gives support a shared home. The team created a private, invite-based web application where people can coordinate tasks, calendar events, and updates for an individual or family going through a major transition.

Within each village, members join as guests or volunteers and contribute according to the role they want to take on. Volunteers can claim tasks, add events, and respond to changing needs, while coordinators have one place to keep everyone informed. Because villages may include sensitive personal information, the team also made privacy and access control a central part of the product’s design.

For Ila Keshishian, the project had a personal connection. “The program provided structure which allowed me to achieve a project that I have wanted to implement as a real issue my family has been facing.” Building Village also gave the team experience developing a product from its earliest planning stages through implementation, testing, and deployment.

The experience gave the students a chance to apply skills from their coursework in a collaborative setting. Amy Yang reflected, “It was incredibly rewarding to work in a team while creating something that could really help people. … As a new summer graduate, I couldn’t have asked for a better way to end my MCIT career!”

“OpenRx”: Bringing Patient-Specific Context to Medication Safety

Members: Priyal Amin, Rui Guan, Mitesh Patel, and Fan Zhang
Mentor: Andrii Podhornyi (MCIT Online ‘24 Alum)

Medication decisions often require clinicians to pull together information from drug labels, treatment guidelines, interaction databases, patient history, and other sources. The OpenRx team developed a clinical decision-support platform that brings this information together and evaluates medications in the context of factors such as drug interactions, allergies, organ function, pregnancy, age, and other medications a patient takes.

OpenRx also includes an Opioid Safety Trajectory feature that tracks changes in opioid-related risk factors over time, helping clinicians identify concerning patterns earlier. The team designed the platform so clinical findings come from structured evidence and predefined logic, while generative AI is used to explain those findings in accessible language. Each finding remains connected to the evidence behind it, giving clinicians a clear path back to the source information.

For Mitesh Patel, OpenRx connected his healthcare background with the computer science skills he developed through MCIT. “SPARC also changed how I think about OpenRx, moving my focus from simply identifying medication problems toward using technology to identify risks earlier and help clinicians intervene before preventable harm occurs.”

Priyal Amin entered the project with little familiarity with pharmacy or clinical software and gradually found her place through the backend work. Reflecting on the experience, she shared, “Finishing OpenRx proved something to me that no classroom ever did, since I can start from zero in a field I know nothing about and still build something meaningful, as long as I keep asking questions and refuse to quit.”

TechPath: Personalized Roadmaps for Technology Careers

Members: Sandy Feng, Zhenhua (Ellie) Hou, Roy Hung, and Kevin Yang
Mentor: Alberto Rusconi (MCIT Online ‘23 Alum)

Technology roles continue to evolve alongside new frameworks, tools, and advances in AI, which can make it difficult at times for early-career engineers to know which skills to prioritize. The TechPath team developed an AI-powered career platform that helps users compare their current experience with the skills associated with different software roles. Users upload a resume to create an editable technical profile, explore potential career paths, and see where their existing skills align with the requirements of each role.

From there, TechPath builds personalized learning recommendations around a user’s career goals. The platform identifies skill gaps, creates training roadmaps, and connects users with relevant MCIT courses, Coursera offerings, and other learning resources. Its career agent updates its guidance as users add skills, complete courses, or change their target role, allowing the roadmap to evolve alongside their progress.

For Sandy Feng, building TechPath offered a new perspective after years of working as a product manager. “For the first time, those vague concepts revealed themselves in the most direct way. I feel lucky to have matched with my teammates where none of us was far ahead, but each of us brought different strengths. Together, we learned from each other and built something tangible.”

Ellie Hou’s work on the AI career agent also strengthened her approach to collaboration and project scope. “Beyond code, I improved at git/PR workflow, coordinating with teammates on parallel work, and scoping what’s actually worth building versus deferring.”

ICD-GPT: Support for Medical Coding

Members: Samuel Cooley, Avery Guan, Xuan Li, and Patricia Vendikos
Mentor: Shreshta Balachandar (MCIT Online ‘23 Alum)

Medical coders translate clinicians’ notes into codes from the standardized system used to classify diagnoses in U.S. healthcare, the International Classification of Diseases, 10th Revision, Clinical Modification (ICD-10-CM). The ICD-GPT team developed an AI-assisted coding tool that reviews clinical notes and proposes relevant codes along with the exact supporting text and a confidence score. Lower-confidence results are routed for human review, keeping coders involved in decisions that require additional judgment.

For Sam Cooley, the project also reinforced the importance of documenting decisions as a team. “Working with three teammates of different backgrounds also taught me that a decision nobody wrote down is a decision that gets remade.”

Patricia Vendikos saw the project as an opportunity to bring her MCIT coursework into a real medical application. “This experience truly bridged the gap between technical skills learned in the MCIT program and practical experience needed to compete in today’s workforce.”

ResumePilot: Tailoring Resumes With AI

Members: Layla Fan, Delphia Li, Isaac Murong, and Weiyi Zhao
Mentor: Benjamin Ye (MCIT Online ‘23 Alum)

Tailoring a resume for different roles requires applicants to understand which job requirements their experience supports and where important gaps remain. ResumePilot gives users a structured way to work through that process. After uploading a resume and adding a job description, users receive an alignment score, a breakdown of matched and missing requirements, and an evidence map showing which parts of the resume support each result.

Users can then generate a tailored resume draft based on their existing experience. ResumePilot keeps the original resume as its source of truth and flags requirements that lack supporting evidence, helping users review the draft before editing, copying, or saving it. 

As the project evolved, the team repeatedly refined the product around the user experience. Delphia Li reflected, “This project was a good example of why software development is iterative.” 

For Weiyi Zhao, ResumePilot provided experience breaking a complex product into manageable pieces. “Working on ResumePilot helped me think more carefully about how to turn an open-ended problem into a sequence of smaller engineering tasks.”

AlignSpace: Turning Renovation Ideas Into Clear Design Decisions

Members: Laura Chen, Sam Effiom, JJ Kim, and Adam Tschida
Mentor: Zere Kairbekova (MCIT Online ‘23 Alum)

Home renovation often starts with scattered inspiration, competing priorities, and difficult tradeoffs around style, function, and budget. AlignSpace helps homeowners organize those inputs into a clearer design direction. The team created an AI-assisted renovation platform that guides users through their preferences, priorities, and budget before producing structured recommendations a professional designer can review.

The experience moves in stages, allowing users to narrow their options before committing to a full renovation package. After gathering information about the room, budget, priorities, and style preferences, AlignSpace ranks possible design directions. Once the user selects one, the system assembles materials, checks the projected budget, suggests alternatives when needed, and produces a brief that carries those decisions into the designer’s handoff.

Building those pieces into one working product also gave the team experience solving problems across different parts of the system. JJ Kim reflected, “Most of our worst bugs lived in the gaps between my code and my teammates’ code, so I got way better at over-communicating, keeping our API contracts in sync, and jumping into Slack to debug together instead of assuming someone else’s part was broken.”

Laura Chen also valued her mentor’s approach to reviewing technical decisions. “Rather than simply giving us answers, they encouraged us to justify our design decisions and consider different implementation approaches.”

Adaptive Response Engine (ARE): Matching AI Confidence to Real-World Risks

Members: Jessie Anderson, Kevin Fickert, Thomas Hefner, and David Nai
Mentor: MCIT Online ‘25 Alum

AI systems sometimes deliver incorrect information with a level of confidence that makes the answer appear reliable. The Adaptive Response Engine (ARE) team explored how AI responses could better reflect both uncertainty and the potential consequences of getting an answer wrong. Their system evaluates each request based on its level of risk and separately assesses the consistency of the model’s response, applying stricter standards to higher-stakes topics such as health and safety.

Based on those signals, ARE decides how an answer should reach the user. Responses with sufficient support are delivered normally, while others include clearer indications of uncertainty, direct users to qualified professionals, or are refused when the request threatens the system’s integrity. 

For Kevin Fickert, building ARE provided a closer look at the limitations of large language models. “From a conceptual standpoint, I’ve learned how LLMs process data, why hallucinations exist, and ways to detect them.”

Thomas Hefner entered the project from a non-engineering background and gained his first experience developing and testing a larger technical system. “Coming from a non-engineering background, SPARC has been my first real experience building and testing a production-style system end to end, and it’s given me a much clearer sense of what day-to-day engineering work actually involves.”

AdvAIsor: AI-Assisted Course Planning for Penn Engineering Students

Members: Sara Bautista, Beth Singer, and Andrew Shugan
Mentor: Connor Hause (MCIT Online ’23 and MSE-DS Online ’25 Alum)

To make course planning resources even more accessible, the AdvAIsor team developed an AI-powered tool that brings together university course information, peer-written reviews, and scheduling data in one place. Students can explore course options based on factors such as workload, difficulty, available study time, degree requirements, and academic goals.

Students begin by sharing their degree program, completed courses, career goals, desired skills, and available study time. AdvAIsor uses that profile alongside course requirements and review data to suggest up to three courses and explain the reasoning behind each recommendation. Students can update their profile over time, allowing the tool to reflect completed coursework and changing priorities while giving them a quick way to explore possible next steps.

For Beth Singer, SPARC offered a chance to gain practical experience with a team while exploring retrieval-augmented generation (RAG). “I participated in SPARC this summer because I wanted more hands-on experience working with a team! I’m hoping to improve and learn more about RAG chatbots and how they are implemented.”

Sara Bautista also valued the chance to combine different areas of technical work into one project. “Doing so has allowed me to explore other skills in creative pursuit of a final product. … I am excited to work on other projects with my newfound skills!”

Recognition 

SPARC 2026 was made possible through the leadership of Mallory Helbling, Career Advisor, the work of Penn Engineering Online Career Services, the alumni mentors who supported teams throughout the summer, and the judges who contributed their time and technical expertise to reviewing the final projects.

We also thank all of the students who participated in SPARC 2026 for the creativity, collaboration, and commitment they brought to their projects throughout the competition.

This year’s judges included:

Efua Mensah-Brown

MCIT Online ‘23 Alum

Matt Parmett

MCIT Online ‘23 Alumni

Learn More About SPARC

Want to see more of what students have built through SPARC? Check out our overview page of the SPARC program to learn how the competition works, or our piece on the 2025 SPARC projects. 

Current MAS-CS and MCIT students can visit the SPARC page for eligibility details and information about participating in a future competition.

Interested in building your computer science skills through Penn Engineering Online? Explore the online Master of Applied Science in Computer Science (MAS-CS) to learn more about the curriculum, student experience, and admissions process.