Datasets:

datamatastudios
/

skill-scarcity-index

snapshot_date stringdate 2026-07-12 00:00:00 2026-08-05 00:00:00	category stringclasses 6 values	skill_name stringclasses 145 values	demand_count int64 10 3.42k	demand_pct float64 0.1 82.3	median_days_open float64 0 108 ⌀	salary_premium_pct float64 -71.1 224 ⌀	repost_rate_pct float64 0 100	scarcity_score float64 0 100
2026-07-12	ai	A/B Testing	115	4.7	1	3.9	2	49.2
2026-07-12	ai	Agile / Scrum	63	2.6	1	42.5	14	62.9
2026-07-12	ai	AI Agents	153	6.3	null	45.5	5.9	72
2026-07-12	ai	AI Coding Tools	32	1.3	null	null	0	0
2026-07-12	ai	Airflow	54	2.2	9	-38.7	16.2	57.2
2026-07-12	ai	AWS	259	10.6	0	-29.2	11.2	31
2026-07-12	ai	Azure	221	9	0	-46.9	19.5	28.3
2026-07-12	ai	BigQuery	43	1.8	4	null	22.2	75.5
2026-07-12	ai	C#	42	1.7	0	-31.5	7.2	25
2026-07-12	ai	C++	66	2.7	1	-10.9	18	60.4
2026-07-12	ai	CI/CD	65	2.7	1	-29.2	17.4	51.8
2026-07-12	ai	Data Modeling	50	2	0	6.8	32.2	54.1
2026-07-12	ai	Databricks	76	3.1	5	-31.5	16.1	58
2026-07-12	ai	Datadog	31	1.3	82	null	0	61.5
2026-07-12	ai	Deep Learning	154	6.3	0	2.7	6.8	37
2026-07-12	ai	Docker	79	3.2	0	-38.7	21	32.2
2026-07-12	ai	ETL	17	0.7	68	-57.5	0	40.1
2026-07-12	ai	Excel	49	2	5	27.2	3.3	64.9
2026-07-12	ai	Fine-tuning	141	5.8	6	-66.8	5.9	36.8
2026-07-12	ai	GCP	190	7.8	1	-53.1	17.3	38.2
2026-07-12	ai	Git	51	2.1	2	-15.1	19.3	67.9
2026-07-12	ai	Grafana	12	0.5	9	41.5	1.6	69.9
2026-07-12	ai	Hugging Face	88	3.6	8	-33	10.6	54.8
2026-07-12	ai	Incident Response	17	0.7	null	null	0	0
2026-07-12	ai	Java	101	4.1	9	-35.3	4.1	50.6
2026-07-12	ai	JavaScript	23	0.9	0	-34.1	29.8	37.4
2026-07-12	ai	Kafka	45	1.8	0	-31.5	1.1	20.9
2026-07-12	ai	Kubernetes	103	4.2	1	-29.2	21.3	55.9
2026-07-12	ai	LangChain	68	2.8	12	-25.6	13.5	72.3
2026-07-12	ai	LLMs / GenAI	861	35.2	0	-17.3	11.3	37.6
2026-07-12	ai	Machine Learning	1,424	58.3	0	-25.6	10.6	34.8
2026-07-12	ai	MLflow	87	3.6	8	-55.4	14.9	47.6
2026-07-12	ai	MongoDB	51	2.1	0	null	16.4	25.8
2026-07-12	ai	Next.js	12	0.5	2	null	0	34.1
2026-07-12	ai	NLP	149	6.1	0	-57.2	10.3	14.2
2026-07-12	ai	Node.js	56	2.3	0	15.2	13.4	44.6
2026-07-12	ai	Pandas	25	1	9	null	21.2	82
2026-07-12	ai	PostgreSQL	52	2.1	2	-29.2	9.4	51.1
2026-07-12	ai	Prompt Engineering	45	1.8	null	null	0	0
2026-07-12	ai	Prototyping	28	1.1	2	null	0	34.1
2026-07-12	ai	Python	506	20.7	0	-42.6	16.9	27
2026-07-12	ai	PyTorch	215	8.8	1	-46.9	15.8	36.9
2026-07-12	ai	RAG	172	7	0	-26.2	13.2	35.7
2026-07-12	ai	React	28	1.1	0	-46.9	7.8	16.8
2026-07-12	ai	Redis	21	0.9	0	null	16.7	26.5
2026-07-12	ai	Redshift	10	0.4	19	-29.2	0	54.4
2026-07-12	ai	Ruby	13	0.5	68	null	0	59.3
2026-07-12	ai	Rust	27	1.1	1	null	0	22
2026-07-12	ai	SageMaker	24	1	0	-31.1	27.5	41.7
2026-07-12	ai	Salesforce	17	0.7	null	null	0	0
2026-07-12	ai	Scala	23	0.9	61	null	0	57.2
2026-07-12	ai	scikit-learn	83	3.4	0	-57.5	35.9	26.5
2026-07-12	ai	Snowflake	16	0.7	19	33.5	0	68.4
2026-07-12	ai	Spark	84	3.4	9	null	10	63.6
2026-07-12	ai	SQL	94	3.8	2	-34.1	14.6	50.1
2026-07-12	ai	Stakeholder Mgmt	143	5.9	0	8.2	4.7	38.1
2026-07-12	ai	Statistical Analysis	101	4.1	8	-36.3	19.6	61.9
2026-07-12	ai	System Design	72	2.9	61	-3.2	19.3	84.5
2026-07-12	ai	TensorFlow	177	7.2	1	-51.8	17	38.5
2026-07-12	ai	Terraform	21	0.9	1	-54.3	3.4	25.9
2026-07-12	ai	TypeScript	17	0.7	1	null	17.7	51.6
2026-07-12	ai	XGBoost	39	1.6	9	-70.3	19.3	50.8
2026-07-12	data	A/B Testing	208	4	16	33.7	2.6	69.1
2026-07-12	data	Agile / Scrum	234	4.5	0	-28.2	21.8	23.2
2026-07-12	data	AI Agents	62	1.2	null	151.2	8.1	69.2
2026-07-12	data	Airflow	147	2.8	9	-1.7	19	71.8
2026-07-12	data	Amplitude	16	0.3	50	-20.9	0	47.7
2026-07-12	data	Apache Iceberg	21	0.4	null	null	0	0
2026-07-12	data	AWS	363	6.9	0	-0.5	18.5	36.1
2026-07-12	data	AWS Security	25	0.5	7	-36	1.3	38.4
2026-07-12	data	Azure	426	8.1	0	-24.4	24.1	25.7
2026-07-12	data	BigQuery	57	1.1	1	45.3	18	67.5
2026-07-12	data	C#	36	0.7	0	-30.2	76.4	28.6
2026-07-12	data	C++	15	0.3	0	-15.1	2.6	20.2
2026-07-12	data	CI/CD	93	1.8	0	10.5	19.9	40.8
2026-07-12	data	ClickHouse	22	0.4	null	null	40.9	96
2026-07-12	data	Dagster	14	0.3	6	null	0	51.6
2026-07-12	data	Data Modeling	340	6.5	0	-18.6	30.4	33.3
2026-07-12	data	Data Pipeline	59	1.1	0	28.4	17	39.4
2026-07-12	data	Data Visualization	87	1.7	0	-18.6	8.2	19.9
2026-07-12	data	Databricks	389	7.4	0	19.1	21.9	44.8
2026-07-12	data	Datadog	14	0.3	35	-25	34.6	65.1
2026-07-12	data	dbt	194	3.7	0	-15.7	15.7	24.7
2026-07-12	data	Deep Learning	31	0.6	0	3.7	12.2	32.5
2026-07-12	data	Docker	42	0.8	1	109.3	15.4	66.9
2026-07-12	data	Elasticsearch	18	0.3	54	151.2	3.2	79.3
2026-07-12	data	ETL	304	5.8	0	-24.4	23.5	25.4
2026-07-12	data	Excel	302	5.8	0	-36	20.6	19.3
2026-07-12	data	Figma	29	0.6	2	null	2.3	54.8
2026-07-12	data	Flink	39	0.7	41	null	15.1	75.3
2026-07-12	data	GCP	88	1.7	0	10.5	19.2	39.8
2026-07-12	data	Git	99	1.9	1	22.4	18.2	65.5
2026-07-12	data	Google Analytics	10	0.2	0	-20.9	18.9	24.9
2026-07-12	data	Grafana	26	0.5	0	112.1	16.7	44.3
2026-07-12	data	GraphQL	23	0.4	1	null	29	71.4
2026-07-12	data	Java	109	2.1	0	110.1	31.2	53.8
2026-07-12	data	JavaScript	23	0.4	0	-2.3	51	43.6
2026-07-12	data	Jira	26	0.5	1	-39.5	31.6	47.7
2026-07-12	data	Kafka	129	2.5	1	-24.4	17.1	43.1
2026-07-12	data	Kotlin	10	0.2	0	151.2	0	33.2

End of preview. Expand in Data Studio

Datamata Skill Scarcity Index

Which tech skills are genuinely hard to hire for: a daily composite scarcity score per skill built from how long roles stay open (time-to-fill), the salary premium employers pay over the category median and how often the same role is re-posted after failing to fill. Computed from active job listings across public company career pages and job boards.

Latest snapshot: 2026-08-05
Rows in this release: 4933
Updated: daily
Licence: CC BY 4.0 — free to use and adapt, including commercially, with attribution.
Source & methodology: https://www.datamatastudios.com/datasets/skill-scarcity-index

Quickstart

import pandas as pd

# Stream straight from the Hub — no download step needed
df = pd.read_csv("hf://datasets/datamatastudios/skill-scarcity-index/skill-scarcity-index.csv")

# Headline: hardest-to-hire data skills right now
latest = df[df.snapshot_date == df.snapshot_date.max()]
print(
    latest[latest.category == "data"]
    .sort_values("scarcity_score", ascending=False)
    [["skill_name", "scarcity_score", "median_days_open", "salary_premium_pct"]]
    .head(10)
)

Or load it with the 🤗 datasets library:

from datasets import load_dataset

ds = load_dataset("datamatastudios/skill-scarcity-index")

What you can answer with it

Which skills keep roles open longest — and whether that gap is widening.
Which skills command a salary premium over their category median.
Which skills employers repeatedly fail to hire for (repost rate).
How scarcity is shifting as AI skills flood into job requirements.

What's NOT in this file (live on the site)

The composite score is here in full. Three deeper cuts are computed from the same pipeline but only published interactively:

Company adoption feed — the first date each company started hiring for each skill ("47 companies added Iceberg to job requirements this quarter"): https://www.datamatastudios.com/datasets/skill-scarcity-index
Stack combinations — which skill triples (e.g. AWS + dbt + Snowflake) actually appear together in postings, monthly.
Time-to-fill by role and seniority — the same lifespan metric cut by role rather than skill.

Columns

Column	Type	Description
`snapshot_date`	string	UTC date the snapshot was taken (YYYY-MM-DD).
`category`	string	Job category: data, engineering, product, devops, security or ai.
`skill_name`	string	Canonical skill name from the extraction taxonomy.
`demand_count`	number	Active listings mentioning the skill on the snapshot date.
`demand_pct`	number	demand_count as a percentage of all active listings in the category.
`median_days_open`	number	Median days recently-closed listings with this skill stayed open. Blank below the sample floor.
`salary_premium_pct`	number	Median disclosed salary of listings with this skill vs the category median, in percent. Blank below the sample floor.
`repost_rate_pct`	number	Share of this skill's listings that are re-posts of an earlier identical role (a failed-hire signal).
`scarcity_score`	number	0-100 weighted percentile-rank composite of the three components within the category. Higher = harder to hire.

How it is built

Each day we snapshot every active job listing scraped from public company career pages and job boards, extract skills with a curated taxonomy and combine three hard-to-hire signals per skill: median lifespan of closed listings (time-to-fill), median disclosed salary vs the category median and the share of listings that are re-posts of an earlier identical role. The score is a weighted percentile-rank composite within each category, so scores are comparable across categories. Full method and known limitations: https://www.datamatastudios.com/methodology.

Citation

Datamata Studios. "Datamata Skill Scarcity Index." 2026-08-05. https://www.datamatastudios.com/datasets/skill-scarcity-index. Licensed under CC BY 4.0.

Downloads last month: 159