Taking the Work Out of Job Hunting

CS major Andrew Allen is developing Pinloop, an AI-powered platform that searches job listings, completes applications and tracks employer responses.

For Andrew Allen, searching for a summer internship meant spending hours moving between job boards, creating accounts and entering information that was already on his résumé. The process left him wondering how much of the work involved in finding a job actually needed to be done by the applicant.

That question led to the development of Pinloop, a platform that uses AI agents to search job listings, recommend positions and complete applications on users' behalf.

Allen (B.S. ’28, computer science), a University of Maryland undergraduate minoring in philosophy, co-founded the startup with Will Marsh (B.S. ’28, applied mathematics), a University of California, Berkeley, undergrad.

Descriptive Image“You just have to open all these different forms, make accounts, fill in the same exact information,” Allen said of the application process. “It’s just pure drudgery, and it’s completely unnecessary for you to have to do that. Oftentimes, it’s the same exact information as it is on your résumé.”

The team is developing the platform through UMD’s Mokhtarzada Hatchery, a program that provides student entrepreneurs with funding, mentorship and workspace to develop their startups.

Frustration to Development

Allen’s interest in using technology to address problems dates back to high school, when he began programming at age 15. Although he initially considered several academic disciplines, his coding experience led him to pursue computer science.

“I just realized that in my classes where I’ve had to code, it felt like despite sometimes having to spend hours on a project, it felt much more rewarding than any other classes that I would take in high school,” Allen said.

After initially pursuing computer engineering at UMD, Allen transferred into computer science, where his interests in programming and software development continued to evolve.

His experience searching for internships eventually gave him an opportunity to apply those interests outside the classroom. During the summer of 2026, Allen began developing a tool that allowed AI agents to search for entry-level computer science positions and internships.

“I wanted to automate that process for myself,” Allen said. “Initially, it started small.”

After sharing the tool on Reddit, Allen received feedback from users interested in opportunities outside computer science and the United States. Their requests prompted him to expand the job database and introduce automated application capabilities.

The Reddit post also led to Allen’s partnership with Marsh, which had been working on similar AI agent applications.

“I saw Andrew’s post on Reddit and thought Pinloop was super cool and interesting because I had been working on agent workflows in a similar way,” Marsh said. “So I emailed him, just a cold email, and we ended up talking for several hours on the phone a few days later. We clicked immediately and got to work.”

Automating the Application

Pinloop launched its web application on Oct. 5, 2026. The subscription-based platform begins by asking users about their professional interests, experience and employment preferences. Users then upload their résumés and provide information employers commonly request.

Through a conversational interface, the system searches a database containing millions of job listings, according to Allen, and evaluates positions against each user’s profile. Users can then authorize the AI agent to complete and submit applications.

The platform uses AI models from OpenAI and Anthropic to evaluate jobs, navigate application forms and generate responses to written questions. It also incorporates technology that identifies similarities between job descriptions and applicants’ professional backgrounds rather than relying solely on keyword matches.

“A ton of windows open in parallel for these job applications that are getting filled out,” Allen said. “You can look at the AI actually clicking around the forms, filling in your information and then submitting it.”

While automating applications is central to Pinloop’s functionality, determining which positions to recommend remains a technical challenge. The team initially relied on an external provider for job listings but is now developing its own system to collect and organize employment data.

“Actually aggregating all the jobs in one place and giving the AI the right tools to search millions of jobs effectively is a difficult problem,” Allen said.

Another challenge is developing AI agents that can navigate application forms that vary across employers. Some require applicants to create accounts, verify their email addresses or complete security checks before submitting applications.

To address email verification requirements, Pinloop allows users to connect their email accounts to the platform, enabling its AI agents to retrieve and enter verification codes.

The team also intends to collect information about employer responses and application outcomes. Those records could help identify which employers and application strategies are associated with higher response rates.

Pinloop had recorded nearly 3,000 sign-ups by Nov. 10. As of Nov. 17, the platform had 183 paying users and was generating over $4,000 in monthly recurring revenue, up from $3,700 a week earlier.

Mentorship

As Pinloop’s user base has grown, maintaining the platform while developing additional features has required Allen to decide where to direct his time and resources.

Through the Mokhtarzada Hatchery, Allen meets with mentors every two weeks to discuss the startup’s progress, address challenges and establish goals.

“Just always having that in the back of your mind really helps you prioritize concrete results rather than just sitting around and thinking about what you might do,” Allen said of the recurring meetings.

Allen said the mentorship has also helped him determine which aspects of the business require immediate attention, particularly when balancing software development, marketing and user concerns.

Looking ahead, the team plans to make Pinloop’s AI agents more autonomous, allowing them to identify opportunities and submit applications in the background while notifying users of their activity.

“You can focus on your classes, focus on your work, or whatever, and you know that applications are getting sent out, and you get email updates about that,” Allen said. “You don’t have to worry anymore about filling out forms, the long hours or the hard work that goes into it; we handle that for you.”

Over the next six months, Marsh said, the team aims to extend that automation beyond submitting applications to securing interviews for users.

“You can type in what jobs you want at night, and then when you wake up in the morning, you'll already have a full schedule of interviews lined up,” Marsh said.

—Story by Samuel Malede Zewdu, CS Communications

The Department welcomes comments, suggestions and corrections.  Send email to editor [-at-] cs [dot] umd [dot] edu.