💡 Inspiration
There is a massive disconnect between the academic theory taught in universities across Bangladesh and the specific, stack-driven skills demanded by Dhaka's rapidly growing tech industry. Many computer science graduates possess strong theoretical foundations but struggle to get hired because they lack targeted experience in the exact frameworks (like React, Spring Boot, or Docker) that top local employers—such as bKash, Pathao, and Brain Station 23—actively list on Bdjobs and LinkedIn. We wanted to build a bridge to close this local skill gap.
🎯 What it does
Dhaka TechMatch is an AI-powered localized career-matching engine. A student simply types in their dream job title (e.g., Full Stack Developer), selects a local ecosystem vertical (like Fintech or Local Software Houses), and pastes their current resume.
The AI instantly:
- Fetches Active Local Jobs: Generates highly realistic, current job openings from real Dhaka companies complete with estimated salary ranges in BDT and exact tech stack requirements.
- Assesses the Skill Gap: Audits the student's resume against these specific local market demands to pinpoint exactly what frameworks or tools they are missing.
- Builds a Roadmap: Generates a strict, hyper-focused 3-month (week-by-week) learning plan with actionable, portfolio-ready project ideas to make them highly hireable in the local market.
⚙️ How we built it
We built the application entirely in Python. For the frontend and routing, we utilized Streamlit for its rapid, reactive web app architecture. The intelligence engine is powered by the OpenAI API (gpt-4o-mini), leveraging advanced prompt engineering to synthesize localized job market data. We used Git/GitHub for version control and deployed the live application securely via Streamlit Community Cloud.
🚧 Challenges we ran into
One of our biggest hurdles was environment security and deployment. We accidentally tracked our local .env file containing our private API keys with Git, which triggered a security block from GitHub. We had to learn how to clear our Git cache, properly configure our .gitignore, and safely inject our secrets via Streamlit's cloud dashboard to get the app live safely. We also had to troubleshoot Streamlit's WebSocket connection issues to get the UI rendering smoothly in production.
🏆 Accomplishments that we're proud of
We went from a blank folder to a fully deployed, AI-powered web application solving a real, localized problem in under 24 hours. We are incredibly proud of the customized prompt engineering that forces the AI to output highly accurate local salary data and company-specific tech stacks.
📚 What we learned
- How to securely manage environment variables (
.env) and prevent API key leaks using.gitignore. - How to deploy a Python app live to the internet using Streamlit Community Cloud.
- Advanced prompt engineering to force an LLM to contextualize its output to a very specific geographical and economic market (Dhaka).
🚀 What's next for Dhaka TechMatch
In the future, we want to integrate live web-scraping directly into Bdjobs and LinkedIn APIs to pull the job listings in real-time. We also want to add user authentication so students can log in, track their 3-month roadmap progress, and check off their weekly milestones!
Log in or sign up for Devpost to join the conversation.