README.md

---
license: odbl
pretty_name: "Internships: Job Postings Snapshot"
task_categories:
  - text-classification
  - token-classification
  - feature-extraction
size_categories:
  - 100K<n<1M
tags:
  - jobs
  - job-postings
  - internships
  - internship
  - co-op
  - early-career
  - students
  - labor-market
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/*.parquet
---

# Internships: Job Postings Snapshot

Internship and co-op job postings. 150,057 jobs, one row per job, from a snapshot taken on 2026-10-19 of the jobs Headcount Labs reads from employers' own hiring systems. 1.8% of the 8,367,419 jobs with a title in the snapshot match.

Source: Headcount Labs (headcountlabs.com), licensed under ODbL 1.0. Live data is at [headcountlabs.com](https://headcountlabs.com).

## Load it

```python
from datasets import load_dataset

ds = load_dataset("HeadcountLabs/internships", split="train")
print(ds.num_rows, ds[0]["title"], "|", ds[0]["employer_name"])
```

To read it without downloading everything, pass `streaming=True` to `load_dataset`.

## Which jobs are included

A job is in this dataset when its `employment_type` (any of the postings merged into the job) contains the word intern, internship, co-op or Praktikum, or when its title contains one of those words as a whole word. 86,788 jobs match on employment type; 63,269 more match on the title only. Employment type is free text that varies by source, so the rule uses whole-word matching: `\b(intern|interns|internship|internships|co-?op|co-?ops|praktikum|praktikant\w*)\b`. The data has no seniority field.

Each row is one job. Postings of the same opening on different sources are merged into one job, and `posting_url` links to an original listing. The job description text is not included. Pay, location and dates are copied as the employer published them.

## Numbers in this snapshot

Every figure on this card is counted from the Parquet files in `data/` of this repository, on 2026-10-19.

| Measure | Value |
| --- | --- |
| Jobs (rows) | 150,057 |
| Rows marked `current` | 149,932 (99.9%) |
| Distinct employer names | 20,487 |
| Posted between | 2010-07-13 and 2026-11-02 |
| Posted in the 30 days before the snapshot | 72,158 |
| Rows with a published pay range | 44,224 (29.5%) |
| Rows with a location | 135,762 (90.5%) |

Work arrangement: 1.1% of the rows state `remote`, 5.3% `hybrid`, 18.4% `in_person` and 75.2% state no arrangement.

44,224 rows (29.5%) carry a published pay range; 40,467 of them are in USD.

Top employers:

| # | Employer | Jobs | Share |
| --- | --- | --- | --- |
| 1 | CVS Health | 2,657 | 1.8% |
| 2 | WALGREENS | 1,916 | 1.3% |
| 3 | Endeavor IT Solution | 1,641 | 1.1% |
| 4 | Walgreens | 1,268 | 0.8% |
| 5 | TikTok | 1,240 | 0.8% |
| 6 | Bosch Group | 946 | 0.6% |
| 7 | Enterprise Holdings | 769 | 0.5% |
| 8 | 10,000 Black Interns 22/23 | 638 | 0.4% |
| 9 | Marriott | 590 | 0.4% |
| 10 | Bombardier | 486 | 0.3% |
| 11 | Enterprise Mobility | 482 | 0.3% |
| 12 | Kimley-Horn | 477 | 0.3% |
| 13 | Stantec | 461 | 0.3% |
| 14 | Schneider Electric | 443 | 0.3% |
| 15 | Pennsylvania CVS Pharmacy, L.L.C. | 442 | 0.3% |

Top locations (first listed location):

| # | Location | Jobs | Share |
| --- | --- | --- | --- |
| 1 | Indore, MP | 1,943 | 1.3% |
| 2 | New York, US-NY | 1,840 | 1.2% |
| 3 | London | 1,364 | 0.9% |
| 4 | Chicago, US-IL | 1,315 | 0.9% |
| 5 | Paris | 1,184 | 0.8% |
| 6 | Singapore | 1,070 | 0.7% |
| 7 | Houston, US-TX | 896 | 0.6% |
| 8 | Austin, US-TX | 866 | 0.6% |
| 9 | Atlanta, US-GA | 835 | 0.6% |
| 10 | Dallas, US-TX | 726 | 0.5% |

Top countries:

| # | Country | Jobs | Share |
| --- | --- | --- | --- |
| 1 | US | 82,902 | 55.2% |
| 2 | DE | 8,507 | 5.7% |
| 3 | FR | 7,020 | 4.7% |
| 4 | IN | 3,863 | 2.6% |
| 5 | CA | 3,671 | 2.4% |
| 6 | BR | 3,172 | 2.1% |
| 7 | SG | 3,079 | 2.1% |
| 8 | NL | 2,283 | 1.5% |
| 9 | GB | 2,166 | 1.4% |
| 10 | IT | 1,822 | 1.2% |

Most common employment types:

| employment_type | rows |
| --- | --- |
| Full time, Internship | 22,029 |
| Intern | 19,823 |
| Internship | 10,732 |
| Part time, Internship | 8,356 |
| INTERN | 5,622 |
| Full-time | 3,899 |
| Full time | 3,453 |
| Full-Time | 3,017 |

Pay intervals (rows with published pay):

| pay_interval | rows |
| --- | --- |
| hour | 34,100 |
| year | 4,670 |
| unknown | 3,281 |
| month | 2,173 |

## Fields

| Column | Type | Meaning |
| --- | --- | --- |
| job_id | string | Stable Headcount Labs job ID (UUID). The same ID is used in the main dataset. |
| title | string | Job title as the employer wrote it. |
| employer_name | string | Employer name as published. |
| company_id | string | Headcount Labs company ID when the employer is matched to a company record, else null. |
| work_arrangement | string | `remote`, `hybrid` or `in_person` when the posting states it, else null. |
| employment_type | string | Employment type as the source states it (free text, for example `Full time`, `Internship`). Null when not published. |
| currentness | string | `current`, `uncertain` or `not_current`: whether we have recent evidence that the job is still open at the snapshot. |
| location_country, location_region, location_city | string | First listed location. Country is an ISO code, region an ISO 3166-2 code such as `US-TX`. |
| location_count | int | Number of locations listed for the job. |
| locations_json | string (JSON) | All locations, each with a `state` and a `value` holding city, region, country and more. |
| pay_min, pay_max | double | Pay range as published, not converted. Null when the employer did not publish pay. |
| pay_currency, pay_interval | string | Currency (for example `USD`) and interval (`hour`, `year`, `month`, `unknown`) as published. |
| posted_at | string | Earliest posting date of the job, ISO 8601 UTC text. |
| job_first_seen_at, job_updated_at | string | When we first collected the job and when its record last changed. |
| posting_url | string | Link to an original listing of the job on the employer's site or board. |
| application_url | string | Link to apply, when published. |
| source_type, source_class | string | Kind of source the posting came from (for example an ATS name) and its class (`ats`, `employer-site`, `government-exchange` and others). |
| posting_count | int | Number of postings (listings on different sources) merged into the job. |
| has_description | int | 1 when the source published a description. The description text is not in this dataset. |

## Source

The jobs are read from employers' public hiring systems: applicant tracking systems, company careers pages and public government job exchanges. Only postings whose source terms allow redistribution are included. Titles, employer names and pay are the employers' own words and numbers. This dataset is a cut of the full Headcount Labs snapshot (the full dataset is coming soon), built from the same export on 2026-10-19. It is a one-off snapshot and is not updated.

## Related

- Full dataset: coming soon. Other public subsets: [huggingface.co/HeadcountLabs](https://huggingface.co/HeadcountLabs)
- Open jobs lists on GitHub, refreshed daily: [internships](https://github.com/HeadcountLabs/internships) · [software-engineering-internships](https://github.com/HeadcountLabs/software-engineering-internships) · [engineering-internships](https://github.com/HeadcountLabs/engineering-internships) · [data-science-ai-internships](https://github.com/HeadcountLabs/data-science-ai-internships) · [business-finance-internships](https://github.com/HeadcountLabs/business-finance-internships)
- Jobs API: [https://headcountlabs.com/docs/](https://headcountlabs.com/docs/)
- MCP server for AI assistants: [https://headcountlabs.com/mcp/](https://headcountlabs.com/mcp/)
- Headcount Labs: [https://headcountlabs.com](https://headcountlabs.com)

## License

[Open Database License (ODbL) 1.0](https://opendatacommons.org/licenses/odbl/1-0/). Our own values in it are under the [Database Contents License (DbCL) 1.0](https://opendatacommons.org/licenses/dbcl/1-0/). Source: Headcount Labs (headcountlabs.com), licensed under ODbL 1.0. Job description text is the employer's own text and is not licensed by us. Takedown requests: https://headcountlabs.com/remove/. Location data includes data from OpenStreetMap contributors (ODbL) and Overture Maps.