Finding a good internship shouldn't feel like finding a needle in a haystack.
My friends and I — a mix of students from
University of California Irvine and
Arizona State University teamed up to solve a problem we were all facing: the lack of a
centralized, accessible, and relevant job board for software engineering internships.
Thus, Techtern Insight was born — a web-based application designed to scrape, store, and display high-quality internship postings tailored to students. It’s not just a job board; it’s a purpose-built tool made by students, for students.
Live site here!Idea
It started with our growing struggle with one thing: finding internships that
weren’t either
outdated, super vague, or totally unrelated to what we were looking for. Every job board felt like a cluttered mess — and we knew we could do better.
We wanted to fix that.
So we asked ourselves:
- What if students had a tool that only showed the roles they cared about?
- What if we could scrape relevant job data and actually make sense of it?
- What if you could see trends like average pay, top cities, and top companies — without digging through a hundred tabs?
Tech Stack & Architecture
- Frontend: HTML, CSS, JavaScript
- Backend: Python Flask
- Database: SQLite3 + MySQL
- Web Scraping: BeautifulSoup4 + Selenium
- Data Visualization: Plotly, Seaborn, and Matplotlib
- Machine Learning: Salary prediction using Linear Regression, Lasso, and Random Forest
Our backend Flask app fetches data from a MySQL database populated via scheduled scrapers. The site's datasite page dynamically renders job listings, and stats provides interactive salary and location-based insights through rich visualizations.
Interactive Visual Insights
We believe data should speak. Users can explore:
- Salary distribution across states
- Top-paying cities and companies
- Density plots and scatterplots of internship pay
- A predictive model to estimate salary ranges based on job attributes
Machine Learning for Pay Prediction
We built salary prediction models using cleaned and encoded datasets.
These models achieved strong R² values and provided users with estimated pay data even when some listings lacked full salary details.
Models used:
- Linear Regression
- Lasso Regression
- Random Forest Regression
This added analytical depth to our platform, making Techtern not just a job board — but a smart one.
Challenges We Tackled
- Scraping dynamic web content (Glassdoor) with strict anti-bot measures
- Transforming inconsistent salary data (hourly, yearly, K-format) into usable numbers.
- Creating a responsive and minimal front end from scratch.
- Hosting and maintaining a stable Flask site with scheduled scraping tasks.
- Integrating ML models for salary prediction within the constraints of a student-led project.
Meet the Team
Sushant Gupta - Tech LeadShanni W. - Fullstack DeveloperDerek Xu - Backend DeveloperFay Alrumaihi - ML Model DeveloperAarav Jain (me) - Site Manager & Deployment Specialist
Together, we iterated through bugs, design pivots, and late-night debugging sessions — learning as much about teamwork as we did about tech.
Final Thoughts
Techtern Insight was more than a summer side project — it was a collaborative solution to a shared pain point. It showcases our ability to build, deploy, and iterate on a product with real user impact. If you're a recruiter or collaborator interested in smart, data-driven job platforms — let's connect!