Here is the complete Devpost/Hackathon submission text, formatted for the specific sections you requested.
Inspiration The inspiration for Professional Profiler came from a growing crisis we observed in our own community: The inflation of "Synthetic Seniority."
With the rise of powerful LLMs like ChatGPT and Copilot, a junior developer—or even a non-coder—can now generate a "Senior Level" GitHub portfolio in a single afternoon. They can flood repositories with AI-generated code, fork complex projects, and manipulate contribution graphs to look like experts.
We realized that recruiters, who spend an average of 6 seconds scanning a resume, are completely ill-equipped to verify this. They are drowning in noise, unable to distinguish between a "Prompter" (who copies AI output) and a "Builder" (who understands architecture). We, Team AlphaCoders, built this tool to bring trust, integrity, and verification back to the hiring process.
What it does Professional Profiler is a "Credit Score for Developer Authenticity." It is a full-stack forensic intelligence tool that audits GitHub profiles to separate real skills from AI noise.
Authenticity Forensics: It calculates a "Trust Score" (0-100) by detecting suspicious patterns:
Burstiness: Does the user upload 6 months of work in one night? (Resume padding).
Fork Ratios: Is 90% of their code just copied from others?
Boilerplate Detection: Does the user rely on tutorial names like my-first-app?
AI Psychological Profiling: It uses Google Gemini 2.5 to read the actual code structure and READMEs. It ignores the "noise" and generates a human-readable Recruiter Summary that explains the developer's actual intent and understanding.
Skill Strength Analysis: It moves beyond keyword matching to determine if a developer is truly "Full Stack" or just "Frontend Focused," providing a SWOT analysis for career growth.
How we built it We architected a Dual-Engine System to handle both quantitative data and qualitative analysis:
The Frontend (React + Tailwind): We built a responsive, "Cyber-Security" themed dashboard. We avoided heavy charting libraries, opting to build custom SVG visualization components from scratch for maximum performance and zero dependencies.
The Heuristic Engine (Client-Side Logic): This mathematical layer interacts with the GitHub API. It fetches repository metadata and runs our proprietary algorithms to calculate "Commit Consistency" and "Originality Scores" in real-time.
The Neural Engine (Gemini 2.5 Flash): This is the brain. We feed the forensic data and top repository descriptions into Gemini. We utilized advanced prompt engineering to force Gemini into the role of a "Skeptical Senior Recruiter," ensuring the output is critical and analytical rather than just a generic summary.
Challenges we ran into GitHub API Rate Limits: Analyzing a user with 50+ repositories triggers rate limits almost instantly. We had to implement a "Smart Triage" system that fetches lightweight metadata first, identifies the top 5 most impactful repositories, and only performs deep AI analysis on those.
Defining "Fake": It is philosophically hard to define what "fake" code is. We didn't want to penalize genuine developers who just happen to code sporadically. We had to tune our algorithm to weigh Complexity (detected by Gemini) higher than just Frequency, ensuring a senior dev who codes once a week isn't flagged as a bot.
Prompt Engineering: Initially, Gemini was too "nice" and would praise even empty repositories. We had to iterate on the system instructions to make it rigorously identify red flags and boilerplate patterns.
Accomplishments that we're proud of The "Forensic Algorithm": We successfully engineered a logic flow that can detect "Sunday Night Dumps"—a common tactic where students upload an entire bootcamp's worth of code in one hour to look busy.
Zero-Server Architecture: We managed to build the entire application as a client-side React app that communicates directly with APIs, making it lightweight, fast, and easy to deploy without managing a Node.js backend.
The UI/UX: We achieved a professional, data-rich interface that looks like a forensic tool, making complex data intuitive for non-technical recruiters.
What we learned Data needs Storytelling: Raw GitHub data (commits, languages) is useless to a recruiter. The value lies in the narrative—transforming "JavaScript: 40%" into "Strong expertise in React ecosystems."
AI as a Filter, not a Judge: We learned that AI works best when it assists human decision-making rather than replacing it. Our "Probability Score" empowers the human to look closer, rather than making the final hiring decision automatically.
Gemini's Context Window: We learned the immense value of Gemini's large context window, which allowed us to feed entire README files for analysis, providing much deeper insights than standard summary tools.
What's next for AlphaCoders-Codespire B2B Recruiter Dashboard: We plan to build a "Bulk Upload" feature where a company can upload a CSV of 500 applicants and get a ranked leaderboard of the most authentic candidates in minutes.
Browser Extension: A Chrome extension that injects the "Authenticity Score" directly onto the GitHub website, warning recruiters in real-time as they browse profiles.
Verified Badge: Creating an embeddable "Verified by AlphaCoders" badge that honest developers can place on their portfolios to prove their work is genuine.
Log in or sign up for Devpost to join the conversation.