Sai_22

Posted on May 16

I Built a Telegram-Based AI Career Agent with Hermes Agent

#hermesagentchallenge #devchallenge #agents

Hermes Agent Challenge Submission

🚀 I Built an Autonomous AI Career Agent Using Hermes Agent

This is a submission for the Hermes Agent Challenge

What I Built

I built Hermes Career Agent — an autonomous AI-powered job application system that runs daily, finds real fresher and intern opportunities, generates tailored resumes for each role, and tracks every application in Google Sheets — all without manual effort.

As a fresher actively applying for AI and software roles, I realized how repetitive and exhausting the job hunting process becomes:

scrolling through job portals daily
checking whether roles match your profile
rewriting resumes for every company
tracking applications manually
forgetting where you already applied

Most candidates end up mass applying with the same generic resume, which usually performs poorly with ATS systems.

So I decided to automate the entire workflow using Hermes Agent.

Every day at 10:00 AM IST, Hermes Agent:

Searches real job listings from verified career portals
Filters relevant fresher/intern opportunities
Scores jobs against my resume using ATS keyword matching
Dynamically rewrites my LaTeX resume for every role
Generates a tailored PDF resume
Sends everything to me on Telegram for approval
Applies automatically after approval
Updates a Google Sheet tracker with all application details

The most interesting part is that the system doesn't just scrape jobs — it actually understands job descriptions and restructures my resume intelligently based on the role requirements.

For example:

GenAI roles prioritize my LLM and RAG projects
Backend roles emphasize FastAPI and APIs
AI Engineering roles highlight orchestration and agentic workflows

Every application gets a unique optimized resume while maintaining the original professional LaTeX formatting.

Demo

Telegram Workflow

The agent sends me:

company name
role
ATS match score
match reasoning
generated resume PDF
approval workflow

After approval, it automatically applies and updates tracking.

Resume Tailoring & ATS Matching

The agent analyzes the job description, extracts important keywords, and dynamically modifies my LaTeX resume to improve ATS compatibility.

Google Sheets Tracking

Every application is automatically tracked with:

Company Name
Role
Job ID
ATS Score
Resume Version
Application Status
Applied Date

Autonomous Daily Workflow

The complete workflow runs autonomously every morning.

Code

The entire implementation is built around Hermes Agent workflows, skills, and autonomous orchestration.

Core Components

career-agent — orchestrates the complete workflow
ats-matcher — extracts keywords and calculates ATS relevance scores
resume-generator — dynamically modifies LaTeX and generates PDFs
firecrawl-scraper — scrapes real jobs from career pages
telegram-notifier — approval workflow and notifications
google-sheets-tracker — application tracking and status management

Project Structure

~/.hermes/career-agent/
├── resume.tex
├── tailored_resumes/
├── ats_scores.json
├── applied_jobs.json
├── telegram_bot.py
├── workflow.py
└── tracker.py

Repository: (Add your GitHub repository link here)

My Tech Stack

Hermes Agent — orchestration, reasoning, workflow automation
Python — ATS scoring, automation, integrations
Telegram Bot API — notifications and approval flow
Google Sheets API — application tracking
LaTeX + Tectonic — professional resume generation
Firecrawl API — scraping real job listings
Playwright — browser automation and applications
Internshala, Wellfound, company career portals — job sources

How I Used Hermes Agent

Hermes Agent is the brain behind the entire system. Instead of building multiple disconnected scripts, I used Hermes Agent to orchestrate the entire workflow autonomously.

1. Autonomous Daily Execution

I configured Hermes Agent to run every day at 10:00 AM IST using cron-based scheduling.

Once triggered, the agent:

searches jobs
filters relevant openings
performs ATS analysis
generates resumes
sends Telegram notifications
waits for approval
applies automatically
updates Google Sheets

No manual intervention required.

2. Multi-Step Tool Orchestration

This project required chaining together:

web scraping
file modifications
LaTeX compilation
browser automation
Telegram messaging
Google Sheets APIs

Hermes Agent handled all of these tools inside a single workflow seamlessly.

Instead of writing glue code between services, the agent manages the orchestration naturally.

3. Intelligent ATS Matching

One of the biggest problems with job applications is ATS filtering.

I built an ATS scoring system where Hermes Agent:

extracts keywords from job descriptions
compares them with my resume
identifies missing skills
calculates a relevance score

Only jobs above a match threshold are surfaced.

This reduced irrelevant applications significantly.

4. Dynamic Resume Tailoring

This is the feature I’m most excited about.

I provided the agent with my original LaTeX resume template.

For every role, Hermes Agent:

analyzes the job description
identifies relevant technologies
rewrites bullet points
prioritizes matching projects
adjusts skill ordering
generates a fresh PDF resume

The result is a unique ATS-optimized resume for every application while preserving the original formatting and structure.

5. Telegram Approval Workflow

Before applying, the agent sends me:

company name
role
ATS score
job link
reasoning
tailored resume PDF

I can simply reply:

apply 1

apply all

The agent then proceeds with the applications automatically.

This approval layer keeps the system autonomous while still giving me control.

6. Automatic Application Tracking

Every application is logged automatically into Google Sheets with:

company
role
job link
ATS score
resume version
status
application date

I also added dropdown-based status management:

Applied
Rejected
Interview
OA
In Progress
Ghosted

This removed the need for maintaining spreadsheets manually.

Why Hermes Agent Was the Right Fit

Traditional automation tools can automate repetitive workflows, but they struggle with reasoning-heavy tasks like:

understanding job descriptions
selecting relevant projects
modifying resumes contextually
making application decisions

Hermes Agent made this possible because it combines:

reasoning
planning
tool usage
file manipulation
autonomous execution

The agent behaves less like a chatbot and more like an autonomous workflow assistant.

This project made me realize that AI agents are not just conversational interfaces anymore — they can handle real-world workflows end-to-end with very little human intervention.

Thanks for reading! 🚀

Building this project was an amazing experience and a great exploration into autonomous AI systems, workflow orchestration, and practical agentic applications.

Built with ❤️ using Hermes Agent by Nous Research.

If you want to reach out to me connect me on linkedin:
https://www.linkedin.com/in/sai-pavan-kumar-devisetti-777553257/

DEV Community