Resum-Aid (POC)

An AI-powered resume enhancement tool that analyzes bullet points, matches roles, and assembles tailored resumes with exportable previews. Built with Bun and Hono, styled with Tailwind CSS, and integrates LLMs via Ollama Cloud or OpenAI.

Note: This repository contains the Proof-of-Concept (POC) and MVP code for the Resume-Aid project. The project has since evolved into a production-grade Career Suite with a private codebase.

🚀 Evolution to Production

While this repository serves as the foundational logic, the Live Production Site has been expanded with enterprise-level features including:

• LinkedIn Profile Builder: A new core module that pivoted the tool from resume-only to a full career optimization suite.
• Production Safety: Implemented fail-fast process-level error handlers (unhandledRejection, uncaughtException) to ensure high uptime and self-healing through Docker orchestration.
• Data Integrity: Added strict file validation and sanitization for secure PDF/Docx processing.
• CRM Integration: Full integration with HubSpot for automated lead management and user tracking.
• Observability: Centralized logging for real-time monitoring of AI generations and system health.

Features (Original POC)

• Bullet Analyzer: Evaluate and improve resume bullets with streaming feedback.
• Job Matcher: Compare resume content against job descriptions for fit.
• Resume Builder: Upload, identify gaps, edit, preview, and download PDF.
• Health checks: GET /ping for liveness.

Tech Stack

• Runtime: Bun (TypeScript)
• Web framework: Hono (server-side routes + views)
• UI/Styles: Tailwind CSS (CLI), minimal custom components
• LLM integration: Ollama Cloud or OpenAI (env-configurable)
• Data & parsing: Mammoth (DOCX → text), unpdf (PDF parsing), jsPDF (PDF generation)
• HTTP & validation: Axios, Zod
• Dev & tooling: Docker Compose, Prettier, Concurrently
• Deployment: Fly.io (fly.toml)
• Testing: Bun test

Screens/Pages

• Landing: src/views/pages/landing.view.ts → /
• Job Matcher: src/views/pages/job-matcher.view.ts → /tools/job-matcher
• Bullet Analyzer: src/views/pages/bullet-analyzer.view.ts → /tools/bullet-analyzer
• Resume Builder: src/views/pages/resume-* → /tools/resume-builder

Quick Start

Local Development (no containers, local Ollama)

bun install
export LLM_PROVIDER=ollama
export OLLAMA_HOST=http://localhost:11434
export OLLAMA_MODEL=gpt-oss:120b-cloud
bun run dev

Open http://localhost:4000.

Cloud Containers (Docker Compose)

# Choose provider (Ollama Cloud)
export LLM_PROVIDER=ollama
export OLLAMA_HOST=https://ollama.com
export OLLAMA_API_KEY=oclk-...
export OLLAMA_MODEL=gpt-oss:120b-cloud

# Build and run
docker compose build
docker compose up -d

# Health check
curl http://localhost:4000/ping

Using OpenAI:

export LLM_PROVIDER=openai
export OPENAI_API_KEY=sk-...
export OPENAI_MODEL=gpt-4o-mini
docker compose up -d

Configuration

• Core: LLM_PROVIDER (ollama | openai, default ollama), PORT (default 4000)
• Ollama: OLLAMA_HOST, OLLAMA_MODEL (default gpt-oss:120b-cloud), OLLAMA_API_KEY (Cloud only)
• OpenAI: OPENAI_API_KEY, OPENAI_MODEL (default gpt-4o-mini)

API Endpoints

• GET /ping: health check
• Tool pages served under /tools/* (job matcher, bullet analyzer, resume builder)
• Resume builder: preview and download endpoints under /api/resume/*

Testing

bun test
bun run test:watch

Deployment

• Fly.io app configured via fly.toml.
• Container exposes 4000 and runs Bun server.

Notes

• Tailwind CSS builds to public/style.css during image build.
• For local Ollama, run without containers.