ApplyBridge

Ctrl+V your resume. Ctrl+Win your dream job.

Comment

ApplyBridge — Devpost Submission Story

Devpost Ctrl+V Hackathon: Making Access Easier
A local-first career companion that helps candidates improve their resumes, manage applications, track recruiter emails, and never miss an important opportunity.

Inspiration

When I started applying for internships in college, I didn't have a mentor, an elder sibling in tech, or an alumni network to show me the unwritten rules of corporate recruiting. Coming from a Tier 3 college, the barriers to entry felt invisible yet insurmountable. I had to teach myself what an "ATS" even stood for, decipher corporate jargon, and figure out how to frame my raw coding projects into language that hiring managers actually respected.

My daily routine became demoralizing: apply for dozens of roles, write customized resumes until 3:00 AM, and then wait in deafening silence. Days bled into weeks with zero responses. To cope with the anxiety, I found myself compulsively refreshing my Gmail inbox, scrolling through hundreds of newsletters and spam, terrified that a recruiter response might slip through the cracks.

Then came the moment that changed everything.

I had applied for a competitive software engineering internship and heard nothing for 5 or 6 days. Naturally, I assumed I was rejected and moved on to other applications. Nearly 10 days later, while digging through a spam-cluttered folder, I accidentally stumbled across an email from their engineering lead: I had been shortlisted, and they had sent a mandatory coding assessment which should required 6 hours to complete.

The catch? Because I discovered the email so late, I had only 5 hours left to complete a technical challenge designed for 6 hours.

I chugged water, turned off all notifications, and stayed up the entire night coding against a ticking clock. My hands were shaking on the keyboard, not because the code was impossible, but because the gatekeeping had almost cost me my dream opportunity simply because an email was buried.

As the sun came up, a sobering realization hit me: How many thousands of talented students, first-generation candidates, and non-traditional developers lose their chance at upward mobility every single day for the exact same reasons?

  • They don't lack intelligence or passion; they lack the arbitrary corporate vocabulary to bypass automated ATS filters.
  • They suffer from application burnout, re-typing the same 5 essay questions across 80 different company portals.
  • They miss critical 48-hour recruiter windows because their inboxes are noisy and chaotic.
  • And worse, existing "job tracker" extensions harvest their private emails and sell candidate data to third parties.

Ctrl+V shouldn't just be about copy-pasting—it should be about pasting down barriers. I wanted to build an air-gapped, zero-knowledge bridge that makes career access genuinely equal for everyone, regardless of what college they attended or who they know.

Hence, ApplyBridge was born.

What It Does

ApplyBridge is a local-first career acceleration system built around six interconnected features:

1. 🪄 The Experience Translator & PDF Studio (Jargon Decryptor)

First-generation students often write what they actually did: "I built a Discord bot with Python and SQLite for our gaming club to assign tournament roles." Traditional ATS parsers reject this as non-commercial.

  • Candidates can upload their PDF resume (parsed 100% in-browser) or paste raw coursework descriptions.
  • ApplyBridge synthesizes 3 executive, recruiter-calibrated bullet variants formatted using the STAR methodology (Situation, Task, Action, Result):
    • Option A (Impact & Metrics-Driven): Highlights scale, throughput, and quantitative percentages.
    • Option B (Technical Depth & Systems Architecture): Focuses on concurrency, caching, data structures, and latency.
    • Option C (Leadership & Ownership): Showcases stakeholder empathy, code reviews, and WCAG AA accessibility standards.
  • Includes a 1-click [Ctrl+V] copy pill to immediately transfer the calibrated bullet into the candidate's resume or application form.

2. The Recruiter Action Radar & Gmail PKCE Protocol

Never miss an interview invitation or short deadline again.

  • Connects directly to Google Identity Services using OAuth 2.0 PKCE in the browser.
  • Strictly Bounded Whitelist Queries: ApplyBridge queries Gmail only for email addresses belonging to recruiters on the candidate’s active pipeline cards (from:[email protected] OR to:[email protected]). Personal, banking, and confidential emails are never read or accessed.
  • Automatically extracts buried action items ("Schedule 45-min screen within 48h"), flags urgency, and generates a pre-drafted polite reply ready to paste.

3. Interactive In-App Interview Scheduler (Zero 404 Links)

When recruiters send third-party Calendly or scheduling links, they frequently expire, hit 404 errors, or conflict with timezones.

  • Candidates select their availability directly within the ApplyBridge modal.
  • Locking in the slot automatically updates the candidate's pipeline status, logs the confirmation in their communication history, exports a one-click Google Calendar event, and fires celebration confetti.

4. The Ghosting Sentinel & Polite Nudge Engine

When a company goes silent for 7+ days, anxiety skyrockets.

  • The Ghosting Sentinel tracks days since last contact and calculates ghosting risk levels.
  • Provides a 1-click polite follow-up nudge generator that drafts professional, recruiter-calibrated check-in messages personalized to the company and role.

5. The Ctrl+V Smart Clipboard Dock

Application fatigue kills momentum.

  • A floating contextual dock that stores reusable answers to perennial screening questions ("Why us?", "Describe a difficult technical bug", "Elevator Pitch").
  • Dynamically interpolates the target company's name, engineering values, and tech stack into the text with instant single-click copying.

6. Air-Gapped Web Crypto AES-256 Vault

  • Job applications contain sensitive Personal Identifiable Information (PII): legal names, phone numbers, home addresses, and confidential emails.
  • ApplyBridge follows a local-first architecture, keeping application data in the browser rather than maintaining a centralized application database. Sensitive vault information is protected using AES-256-GCM encryption.

How We Built It

ApplyBridge is built from the ground up as a high-craft, production-grade web application combining functional minimalism with a bold Neo-Brutalist design language.

┌────────────────────────────────────────────────────────────────────────┐
│                          APPLYBRIDGE CLIENT                            │
│                                                                        │
│  ┌───────────────────────┐  ┌─────────────────┐  ┌──────────────────┐  │
│  │ Neo-Brutalist UI      │  │ Experience      │  │ Ctrl+V Smart     │  │
│  │ (Bento Grid + Poppins)│  │ Decryptor Studio│  │ Clipboard Dock   │  │
│  └───────────┬───────────┘  └────────┬────────┘  └────────┬─────────┘  │
│              │                       │                    │            │
│  ┌───────────▼───────────────────────▼────────────────────▼─────────┐  │
│  │                Universal Multi-Model AI Router                   │  │
│  │  (Google Gemini Flash · Hugging Face API · Groq LPUs · Local)    │  │
│  └───────────────────────────────────┬──────────────────────────────┘  │
│                                      │                                 │
│  ┌───────────────────────────────────▼──────────────────────────────┐  │
│  │                   Client-Side Security Sandbox                   │  │
│  │  • Regex PII Sanitizer & Masker   • Whitelist Gmail Query Engine │  │
│  │  • PDF.js Dynamic ESM Loader      • Web Crypto AES-256-GCM Vault │  │
│  └───────────────────────────────────┬──────────────────────────────┘  │
│                                      │                                 │
│  ┌───────────────────────────────────▼──────────────────────────────┐  │
│  │               Encrypted Local IndexedDB (Dexie.js)               │  │
│  │               (0 Bytes Transmitted to Cloud DBs)                 │  │
│  └──────────────────────────────────────────────────────────────────┘  │
└────────────────────────────────────────────────────────────────────────┘

The Technology Stack:

  • Core Architecture: React 19, TypeScript 5.8, Vite
  • Styling & Design System: Tailwind CSS with custom Neo-Brutalist utility classes (border-3 border-slate-900, shadow-[5px_5px_0px_#0f172a]), Google Font Poppins typography, and Canvas Confetti for celebratory micro-interactions.
  • Local Database & Offline Persistence: Dexie.js reactive IndexedDB with schema migrations for applications, recruiter communications, user profiles, and clipboard snippets.
  • Cryptographic Security Layer: Native browser Web Crypto API implementing AES-256-GCM, PBKDF2 (100,000 iterations), and SHA-256 authentication hashes.
  • Client-side PDF parsing using PDF.js.: Dynamic CDN ESM loader importing Mozilla's PDF.js directly into the client execution thread, extracting text layers on the candidate's CPU without server uploads.
  • Universal Multi-Model AI Routing: Built-in auto-detection layer supporting:
    • Google Gemini 1.5/2.0 Flash: Generous 1,500 reqs/day free tier with a 1M token context window.
    • Hugging Face Serverless Inference API: Open-weights access (meta-llama/Meta-Llama-3-8B-Instruct, mistralai/Mistral-7B-Instruct-v0.2, Qwen/Qwen2.5-72B-Instruct).
    • Groq Cloud LPUs: Ultra-low latency inference.
    • Local Heuristic Synthesis Engine: Zero-latency offline fallback that works with 0 API keys.
  • Gmail Protocol: Client-side Google Identity Services (GIS) OAuth 2.0 PKCE tokens restricted to gmail.readonly.
  • Authentication & Security Rules: Zero-trust Firebase Auth SDK loaded dynamically via Google CDN, enforced by production-grade firestore.rules and storage.rules with strict Default Deny-All policies.

Challenges We Ran Into

1. Using AI on resumes without sending someone's personal details

A resume contains a name, phone number, email and often an address. I didn't want to send that to a cloud LLM, but the model still needs the resume text to rewrite it. I wrote a small scrubber (PIIService.ts) that replaces identifiers with placeholders like [CANDIDATE_NAME] and [PHONE_REDACTED] before the text goes out, then puts the real values back locally when the response returns.

It is regex-based, so it is a best-effort filter, not a guarantee. It handles structured details like phone numbers and emails better than details written in free text, such as a project description that names a specific company or town. Whatever text remains still goes to the AI provider the user picks.

2. Reading a recruiter's emails without reading the rest of the inbox

When a user adds an application with a recruiter's address, the app asks Gmail only for messages matching that address:

from:[email protected] OR to:[email protected]

The catch is that Google's gmail.readonly scope covers the whole mailbox, so this limit comes from how my app builds its queries, not from Google enforcing it. A user has to trust the app on that point. The requests are made from the browser, so they can be inspected in the network tab, and the code is open for anyone to check.

3. Running out of disk space mid-build

Partway through development my laptop hit ENOSPC: no space left on device. The Firebase SDK and PDF.js were taking up a lot of space in node_modules, so I switched both to loading from a CDN at runtime. That solved the disk problem, with a tradeoff: the app now needs network access to load those scripts, so it is not fully offline.

Accomplishments That We're Proud Of

  1. A Real Working Solution for Non-Traditional Students: I created a tool that directly addresses the pain points of students who don't have insider guidance. Watching informal descriptions turn into executive STAR bullets with quantifiable metrics is genuinely empowering.
  2. Local-first Architecture: Unlike commercial scrapers that harvest candidate data, ApplyBridge stores everything locally in the candidate's browser vault with AES-256-GCM encryption.
  3. Neo-Brutalist Design with Soul: I eschewed generic AI templates in favor of a bespoke, tactile aesthetic—bold 3px solid ink borders, electric lime and emerald accents, interactive 7-day velocity charts, and celebratory confetti.
  4. Interactive Hero Decryptor Playground: Anyone visiting the landing page can test the live Experience Translator sandbox in the hero section—with no signup or login required.
  5. *Eliminated interview scheduling errors: * by replacing unreliable third-party links with an integrated, in-app scheduler that automatically exports Google Calendar invites.

What We Learned

  • Language is a Gatekeeper: I learned how deeply recruitment jargon undermines capable students from Tier 3 colleges and underrepresented backgrounds. Framing coursework as "engineered client-side data pipelines" rather than "built a class project" can be the difference between selection and rejection.
  • The Power of Client-Side Web Crypto: Modern browsers possess immense cryptographic capabilities. Getting to know crypto.subtle for AES-256-GCM and PBKDF2 so that applications do not need central cloud backends to deliver enterprise-grade security.
  • The Value of Functional Minimalism: By giving candidates a single unified platform which include Kanban pipeline, recruiter email radar, resume studio, and clipboard dock, I reduced the mental fatigue of job hunting by a miles!

What's Next for ApplyBridge

  • [ ] Chrome Extension Companion: Inject the Ctrl+V Smart Clipboard Dock directly into Workday, Greenhouse, Lever, and LinkedIn forms for 1-click autofill without leaving the job application page.
  • [ ] Community Jargon De-Anonymizer: An open-source, crowdsourced database of verified recruiter rubrics for top tech companies to help first-generation candidates anticipate technical screening criteria.
  • [ ] Real-Time Voice Mock Interviewer: An in-browser AI practice partner calibrated to the specific job description that simulates technical phone screens with instant feedback.
  • [ ] Mobile Progressive Web App (PWA): Offline-first mobile version with push notifications alerting candidates when a recruiter deadline is within 12 hours.

ApplyBridge was built with passion for the Devpost Ctrl+V Hackathon 2026: Making Access Easier.
Live Application: https://apply-bridge-pi.vercel.app/
GitHub Repository: TarequeSyed/Apply-Bridge

Built With

Share this project:

Updates

Submission history