HackerRank's engineering team published an MIT-licensed Python script that scores resumes automatically, enriches them with live GitHub data, and produces an explainable candidate evaluation, the same workflow companies pay $35 to $55 per user per month to get from AI-enabled ATS platforms. For a five-person recruiting team, that is up to $3,300 per year in ATS fees for this one capability, running for free on a laptop.
Hiring Agent, an MIT-licensed Python script published by HackerRank's engineering team, automates the initial resume scoring workflow that companies currently run manually or pay $35 to $55 per user per month to get from AI-enabled ATS platforms. For a five-person recruiting team on Manatal's Enterprise Plus tier, that's $3,300 per year just for AI-assisted candidate evaluation. Hiring Agent runs that specific job for free, on your own hardware, in roughly 30 seconds per resume.
That is not a small number to spend on a feature that amounts to "an AI reads this PDF and tells me if the person is worth a phone call."
What Hiring Agent Actually Does
The workflow is straightforward. You hand the tool a resume PDF. It converts the document to structured text, then uses a local or hosted language model to extract every relevant section: work history, education, skills, awards, and project experience. If the resume includes a GitHub profile link, the tool pulls that account's repository data, classifies each project by type and complexity, and selects the seven most relevant ones as additional evidence.
Then it scores the candidate. The rubric covers four core dimensions: open-source contributions, self-directed projects, production experience, and demonstrated technical skills. Each dimension gets a score backed by supporting evidence pulled directly from the resume and GitHub data. Bonus points and deductions apply based on criteria defined in customizable prompt templates. The final output is a readable report explaining how a candidate arrived at their overall score, not just what the number is.
That explainability is actually the most interesting design choice here. One of the consistent complaints about commercial AI screening tools is that they produce scores without justification, making it difficult to catch bias or explain a decision to a candidate. Hiring Agent's scoring is transparent by design: every score traces back to a specific piece of text or a repository the model found and evaluated. That transparency is easier to audit than a black-box vendor score.
The Setup Reality
Installation is not point-and-click. You need Python 3.11 or later, plus either a local Ollama instance with a model pulled (gemma3:4b works on most laptops, gemma3:12b performs better on stronger machines) or a Google Gemini API key. Setup takes 10 to 15 minutes for someone comfortable in a terminal, and closer to an hour for someone who isn't. Once configured, the command to score a resume is a single line pointing at a PDF path.
The tool outputs to your console and, with development mode enabled, appends rows to a CSV file. There is no web interface, no dashboard, no database. It is a script that produces a text report. For teams that have basic technical capacity, that is workable. For teams who need something they can hand to a non-technical hiring manager with a login, it is not there yet.
What You Give Up
Hiring Agent is not an ATS. It has no job posting, pipeline management, interview scheduling, offer letter workflow, or integrations with any recruiting tool your team already uses. It will not connect to Greenhouse, Lever, or Workday. It does not send emails, log activity, or maintain a searchable candidate database. It scores one resume at a time from the command line.
There is also no compliance framework built in. Regulated industries and companies with formal EEOC review processes should approach this carefully. The scoring criteria are configurable via Jinja prompt templates, which gives you precise control over the rubric, but it also means the defaults reflect HackerRank's institutional model of what a strong technical candidate looks like. For engineering roles, that default rubric is meaningful. For marketing, finance, or operations roles, you will likely need to rewrite the evaluation templates from scratch before the scores reflect anything useful.
The Cost Argument
SHRM puts the average cost per hire at $4,700 as of 2023, and that number has continued to climb. A significant fraction of that cost is recruiter and coordinator time spent on first-pass screening, reading resumes that would have been filtered out within five minutes by anyone with domain knowledge.
For a company hiring 40 to 60 people per year, primarily in technical roles, Hiring Agent represents a legitimate proof of concept for AI-assisted first-pass screening at zero software cost. Even running it on a subset of applications, as a sanity check before a recruiter spends time on a candidate, changes the economics of the screening step.
The caveat is honest: this is a script, not a product. It does not have the reliability guarantees, audit logging, or support tier that a paid vendor provides. But if you are evaluating whether to add AI screening to your hiring process before committing to a vendor contract, Hiring Agent gives you a real workflow to test against your actual resumes.
Who Should Pay Attention
Engineering-led companies doing high-volume technical hiring who have a developer who can set up and maintain the script. HR and talent ops leaders who want to build intuition about AI screening before buying into a platform. Teams that hire in bursts and do not want to pay per-seat ATS fees year-round for a capability they use seasonally.
A Note on Who Built This
What makes this particular release worth paying attention to is not the code. It is the rubric.
HackerRank evaluates developer candidates at a scale few companies reach. Their scoring logic for open-source contributions, project depth, and production signal reflects years of understanding which resume signals actually predict engineering performance. When that institutional knowledge ships as a free MIT-licensed script with configurable templates, any company can inspect exactly how a candidate gets evaluated and adjust those criteria to match their own hiring bar.
That is a different kind of asset than a generic LLM prompt. It is compiled professional judgment, and right now it is a free download.