Apppublic

Omtiwari553/AutoOps-Ai

sourceHugging Faceupdated 6mo agoView on Hugging Face

0likes

App README

AutoOps AI+

Autonomous Career Task & Preparation Environment — OpenEnv Hackathon Submission

![OpenEnv Compatible](https://openenv.dev) ![Python 3.10+](https://python.org) ![FastAPI](https://fastapi.tiangolo.com)

Description

AutoOps AI+ is a fully autonomous career task environment built for the OpenEnv hackathon. It simulates the end-to-end job search workflow — from resume analysis and job discovery to application submission, cover letter generation, interview scheduling, and mock interview practice.

An AI agent interacts with the environment step-by-step using a rich set of career tools. Each action moves the agent closer to its career goal, while a deterministic reward function provides meaningful signal about the quality of its decisions. The environment prevents spam applications, rewards high-quality matches, and penalizes poor performance.

The three task tiers (easy → medium → hard) progressively test the agent's ability to plan multi-step career strategies, balance competing priorities (apply broadly vs. apply well), and demonstrate self-improvement across mock interview sessions.

Observation Space

Field	Type	Description
`current_step`	`int`	Steps taken so far
`max_steps`	`int`	Episode step budget
`user_goal`	`str`	The career goal to accomplish
`available_actions`	`List[str]`	Legal actions at this step
`applications`	`List[Application]`	Jobs applied to
`interviews`	`List[Interview]`	Scheduled interviews
`mock_sessions`	`List[MockSession]`	Practice interview results
`skill_gaps`	`List[str]`	Missing skills identified
`last_action_result`	`str`	Result of previous action
`last_action_success`	`bool`	Whether last action succeeded
`reward_so_far`	`float`	Cumulative reward in episode
`done`	`bool`	Episode complete flag

Action Space

Tool Name	Args	Description
`analyze_resume`	`resume_text: str`	Parse skills from resume text
`search_jobs`	`role, location, limit`	Find matching jobs
`smart_apply`	`job_id: int`	Apply if match score > 0.5
`generate_cover_letter`	`job_id, tone`	Personalized cover letter
`schedule_interview`	`company, date, type`	Book an interview slot
`send_email`	`to, subject, body, type`	Send a career email
`conduct_mock_interview`	`questions, answers`	Practice interview session
`skill_gap_analysis`	`job_skills: list`	Identify missing skills
`salary_research`	`role, location`	Research salary range data
`finish`	—	End the episode

Tasks

Task	Difficulty	Max Steps	Goal
`easy_01`	Easy	10	Apply to 1 job with match score > 0.5
`medium_01`	Medium	20	Apply to 2-5 jobs, schedule interview, send follow-up
`hard_01`	Hard	40	Full pipeline: apply, mock prep, skill gaps, salary research

Reward Function

Event	Reward
Quality application (match ≥ 0.6)	+0.30
Average application (match ≥ 0.5)	+0.15
Interview scheduled	+0.20
Strong mock interview (score ≥ 70%)	+0.30
Average mock interview (score ≥ 50%)	+0.10
Cover letter generated	+0.15
Skill gap identified	+0.10
Salary researched	+0.10
Email sent	+0.05
Spam application (match < 0.5)	-0.20
Duplicate application	-0.15
No improvement after 3 mocks	-0.30
Tool error	-0.05

Setup

bash

# 1. Install uv
pip install uv

# 2. Create virtual env and install deps
uv sync

# 3. Generate uv.lock (REQUIRED for openenv validate)
uv lock

# 4. Start server locally for testing
uv run autoops-server
# OR: uvicorn server.app:app --host 0.0.0.0 --port 7860

Test Endpoints

bash

# Health check
curl http://localhost:7860/health

# Reset environment
curl -X POST http://localhost:7860/reset \
  -H "Content-Type: application/json" -d '{}'

# Get state
curl http://localhost:7860/state

# Take a step
curl -X POST http://localhost:7860/step \
  -H "Content-Type: application/json" \
  -d '{"tool_name": "analyze_resume", "args": {"resume_text": "React, JS, CSS"}}'

# List tasks
curl http://localhost:7860/tasks

Run Inference Agent

bash

export API_BASE_URL="https://router.huggingface.co/v1"
export HF_TOKEN="your_token_here"
export MODEL_NAME="Qwen/Qwen2.5-72B-Instruct"
export SERVER_URL="http://localhost:7860"
python inference.py

Run Tests

bash

uv run pytest tests/ -v

Docker

bash

docker build -t autoops-ai-plus .
docker run -p 7860:7860 \
  -e API_BASE_URL="https://router.huggingface.co/v1" \
  -e HF_TOKEN="your_token_here" \
  -e MODEL_NAME="Qwen/Qwen2.5-72B-Instruct" \
  autoops-ai-plus

Environment Variables

Variable	Description	Default
`API_BASE_URL`	LLM API endpoint	`https://router.huggingface.co/v1`
`MODEL_NAME`	Model identifier	`Qwen/Qwen2.5-72B-Instruct`
`HF_TOKEN`	HuggingFace / API key	—
`SERVER_URL`	Environment server URL	`http://localhost:7860`
`MAX_STEPS`	Max steps per inference episode	`15`

Baseline Scores

Task	Score	Reward	Notes
`easy_01`	0.85	0.45	Apply + quality check
`medium_01`	0.72	0.70	Apply + interview + follow-up
`hard_01`	0.61	0.95	Full pipeline

Project Structure

autoops-ai-plus/
├── README.md
├── Dockerfile
├── pyproject.toml
├── uv.lock
├── openenv.yaml
├── inference.py              ← Root-level OpenAI agent
├── server/
│   ├── __init__.py
│   └── app.py                ← FastAPI server
├── environment/
│   ├── __init__.py
│   ├── env.py                ← CareerEnv (step/reset/state)
│   ├── models.py             ← Pydantic models
│   └── reward.py             ← Reward calculator
├── tasks/
│   ├── __init__.py
│   ├── definitions.py
│   ├── easy_task.json
│   ├── medium_task.json
│   └── hard_task.json
├── graders/
│   ├── __init__.py
│   ├── easy_grader.py
│   ├── medium_grader.py
│   └── hard_grader.py
├── tools/
│   ├── __init__.py
│   ├── resume_tool.py
│   ├── job_search_tool.py
│   ├── apply_tool.py
│   ├── email_tool.py
│   ├── calendar_tool.py
│   ├── interview_tool.py
│   ├── skill_tool.py
│   └── cover_letter_tool.py
└── tests/
    ├── test_env.py
    └── test_graders.py

Author

OmTiwarii — OpenEnv Hackathon 2025