🌌 ASTRA

Autonomous System for Talent & Resume Automation

Local-first AI system for ATS resume optimization, cover letter generation, job-aware chat, and LaTeX-safe document workflows.

📘 Overview

ASTRA is a local-first, modular AI ecosystem built to help candidates generate stronger, more targeted application materials with a desktop-style experience.

It combines:

JD-aware resume optimization
Role-specific cover letter generation
Humanized writing support
Persistent job-aware chat
Safe LaTeX compilation with PDF output
Local history, logs, and reusable session context

ASTRA runs through a FastAPI backend + PyWebView desktop application, giving a native app-like workflow with preview, saved sessions, and structured output generation.

🪐 ASTRA Subsystems

Subsystem	Description
🧠 HIREX	Core resume optimization engine. Parses job descriptions, rewrites LaTeX resumes, improves ATS alignment, and compiles final PDFs.
🗣️ SuperHuman	Humanization layer that rewrites content to sound natural, professional, and recruiter-friendly while staying LaTeX-safe.
💬 MasterMind	Job-aware conversational assistant with persistent memory, context retention, and multi-turn support.

✨ What ASTRA Does

ASTRA is designed to generate better job application outputs, not just generic text.

Core capabilities

Optimize ATS-friendly LaTeX resumes
Generate job-specific cover letters
Provide context-aware Q&A using saved JD and resume context
Maintain persistent local chat sessions
Track history, analytics, and output metadata
Compile documents through a safe LaTeX pipeline

🧩 Main Features

Module	Purpose
🧾 Resume Optimizer	Rewrites LaTeX resumes using JD-aware bullet planning, skill alignment, and one-page fit control.
✍️ Cover Letter Engine	Produces tailored cover letters using company, role, and resume context.
💬 Talk to ASTRA	Lets users ask questions about a saved JD, resume, or generated output.
🗣️ SuperHuman	Humanizes bullets, responses, and cover letters into more natural language.
🧠 MasterMind	Stores and manages conversation context across multiple sessions.
📊 Dashboard	Tracks usage patterns, history, fit trends, and output activity.
⚙️ Utilities / Routers	Handles config, model routing, validation, telemetry, and helper functions.

🧠 Resume Optimizer Highlights

The current optimizer is not a simple keyword replacer. It performs a multi-stage optimization pipeline designed for stronger final resume quality.

Resume optimization pipeline

Extracts must-have, should-have, and nice-to-have skills from the JD
Classifies the role into a role archetype
Decomposes the JD into day-to-day tasks
Extracts JD key phrases that should appear naturally in the resume
Profiles the ideal candidate
Builds a 12-bullet master plan across experience sections
Generates JD-mirrored project entries
Runs post-generation validation and repair
Rebuilds the Skills section using JD-priority ranking
Applies safe one-page trimming
Injects PDF metadata
Compiles the final LaTeX safely into PDF

Output-quality features

Task-aware bullet generation
JD phrase mirroring
Metric diversity enforcement
Placeholder cleanup for bad generations like XYZ, ABC, Foo, Lorem, etc.
Cross-block bullet deduplication
Bullet quality scoring and improvement
Coverage remediation for missing must-have JD terms
Skills prioritization based on JD relevance
One-page preservation rules so experience bullets are not over-trimmed
Projects section generation in \textbf{Title} -- one-liner format

🏗️ Project Structure

ASTRA/
│
├── backend/
│   ├── api/
│   │   ├── optimize.py
│   │   ├── coverletter.py
│   │   ├── talk.py
│   │   ├── superhuman.py
│   │   ├── humanize.py
│   │   ├── mastermind.py
│   │   ├── dashboard.py
│   │   ├── context_store.py
│   │   ├── models_router.py
│   │   ├── utils_router.py
│   │   └── debug.py
│   │
│   ├── core/
│   │   ├── config.py
│   │   ├── compiler.py
│   │   ├── security.py
│   │   └── utils.py
│   │
│   └── data/
│       ├── contexts/
│       ├── history/
│       ├── logs/
│       ├── mastermind_sessions/
│       └── cache/
│
├── frontend/
│   ├── master.html
│   ├── master.js
│   ├── static/css/
│   └── static/assets/
│
├── main.py
└── requirements.txt

⚙️ Setup

1. Install dependencies

pip install -r requirements.txt

If requirements.txt is still incomplete, install the core packages manually:

pip install fastapi uvicorn httpx openai python-dotenv pywebview pydantic

2. Configure environment variables

Create a .env file:

OPENAI_API_KEY=sk-xxxxxx
HUMANIZE_API_KEY=Bearer xxxxx
DEBUG=true
DEFAULT_MODEL=gpt-4o-mini
API_BASE_URL=http://127.0.0.1:8000

You may also configure default file paths and local storage directories inside backend/core/config.py.

3. Run ASTRA

python main.py

Launch behavior

FastAPI backend starts on 127.0.0.1:8000
PyWebView opens the desktop UI
Sessions, logs, and generated artifacts persist locally

Useful endpoints:

http://127.0.0.1:8000
http://127.0.0.1:8000/api/docs

🧾 Backend Modules

`optimize.py` — HIREX Core

Main resume optimization engine.

Responsibilities include:

extracting JD requirements
identifying role type and tone
planning and rewriting experience bullets
generating project entries
rebuilding the skills section
safe LaTeX compilation
one-page fitting logic
resume coverage and alignment reporting

`coverletter.py`

Generates role-specific cover letters using:

JD context
target company and role
existing resume profile
safe output formatting

`talk.py`

Supports contextual Q&A tied to:

the current JD
saved resume context
previous ASTRA sessions
generated outputs

`superhuman.py`

Applies humanization and tone control for:

resume bullets
cover letters
answers
recruiter-facing writing

Designed to keep outputs natural without breaking LaTeX structure.

`mastermind.py`

Provides:

persistent local session storage
multi-turn reasoning
contextual conversational support
reusable job-aware memory

`context_store.py`

Stores reusable context bundles such as:

JD text
selected resume state
generated outputs
prior chat state

`dashboard.py`

Aggregates usage and application-related signals such as:

generation frequency
session activity
optimization history
fit-score or coverage trends

💾 Local Data Storage

Directory	Purpose
`backend/data/logs/events.jsonl`	Event logs
`backend/data/history/history.jsonl`	Output and usage history
`backend/data/contexts/`	Saved JD + resume bundles
`backend/data/mastermind_sessions/`	Persistent chat sessions
`backend/data/cache/latex_builds/`	Temporary LaTeX builds

🔐 Security and Safety

ASTRA uses a defensive document pipeline:

strict .tex validation
secure input filtering
temporary sandboxed LaTeX build directory
no shell escape during compile
controlled local file handling
LaTeX-safe escaping before rendering

This helps reduce broken builds and unsafe LaTeX execution patterns.

📈 Logging and Analytics

ASTRA uses structured event logging for local observability.

Example:

log_event("event_name", {"meta": {...}})

Common event types:

optimize_resume
coverletter_draft
talk_answer
superhuman_rewrite
frontend_debug

Logs are stored in:

backend/data/logs/events.jsonl

🧱 Run Modes

Mode	Command
Full desktop app	`python main.py`
API-only dev mode	`uvicorn backend.api:app --reload`
API docs	`/api/docs`

🚀 Roadmap

Planned upgrades include:

richer fit scoring and recruiter-style evaluation
better session retrieval and memory search
PDF-to-LaTeX conversion
streaming generation
dashboard improvements
stronger artifact versioning
more advanced analytics for resume evolution

🪙 License and Attribution

Educational and research use permitted.

ASTRA™, HIREX™, SuperHuman™, MasterMind™ are associated with the author’s project ecosystem.

👤 Author

Sri Akash Kadali

Applied Machine Learning Graduate Student University of Maryland

Intelligence that understands your profile, aligns your story to the role, and improves resume output quality end to end.

Name		Name	Last commit message	Last commit date
api		api
backend		backend
core		core
data		data
frontend		frontend
latex_errors		latex_errors
.gitignore		.gitignore
LICENSE		LICENSE
README.md		README.md
executable.txt		executable.txt
favicon.ico		favicon.ico
fix_astra.py		fix_astra.py
fix_astra2.py		fix_astra2.py
fix_astra3.py		fix_astra3.py
fix_astra4.py		fix_astra4.py
icon.icns		icon.icns
main.py		main.py
revert_astra.py		revert_astra.py

Navigation Menu

Repository files navigation

🌌 ASTRA

📘 Overview

🪐 ASTRA Subsystems

✨ What ASTRA Does

Core capabilities

🧩 Main Features

🧠 Resume Optimizer Highlights

Resume optimization pipeline

Output-quality features

🏗️ Project Structure

⚙️ Setup

1. Install dependencies

2. Configure environment variables

3. Run ASTRA

Launch behavior

🧾 Backend Modules

`optimize.py` — HIREX Core

`coverletter.py`

`talk.py`

`superhuman.py`

`mastermind.py`

`context_store.py`

`dashboard.py`

💾 Local Data Storage

🔐 Security and Safety

📈 Logging and Analytics

🧱 Run Modes

🚀 Roadmap

🪙 License and Attribution

👤 Author

About

Releases

Packages

Contributors

Languages

Latest commit

History

Folders and files

Repository files navigation

🌌 ASTRA

📘 Overview

🪐 ASTRA Subsystems

✨ What ASTRA Does

Core capabilities

🧩 Main Features

🧠 Resume Optimizer Highlights

Resume optimization pipeline

Output-quality features

🏗️ Project Structure

⚙️ Setup

1. Install dependencies

2. Configure environment variables

3. Run ASTRA

Launch behavior

🧾 Backend Modules

optimize.py — HIREX Core

coverletter.py

talk.py

superhuman.py

mastermind.py

context_store.py

dashboard.py

💾 Local Data Storage

🔐 Security and Safety

📈 Logging and Analytics

🧱 Run Modes

🚀 Roadmap

🪙 License and Attribution

👤 Author

About

Resources

Stars

Watchers

Forks

Releases

Packages

Contributors

Languages

`optimize.py` — HIREX Core

`coverletter.py`

`talk.py`

`superhuman.py`

`mastermind.py`

`context_store.py`

`dashboard.py`