The most in-demand skills in job postings, and how many ask for AI
The skills employers name in their own job postings, counted across the open roles our crawler reads on employers’ careers sites. A model reads each posting and lists the skills it asks for; this page counts them.
- Open roles
- 74,216
- Skills read from
- 16,917
- Ask for an AI skill
- 1,366
- Employers asking for AI
- 270
that we track
23% of open roles so far
roles read so far
among the roles read
How many job postings have we read?
Skills read from 16,917 of the 74,216 open roles we track so far (23%) — under the 50% of open roles and 2,000 roles we need before quoting a percentage, so this page gives counts until the parser catches up.
What are the most in-demand skills in job postings (2026)?
Download CSVSkills named in at least 3 roles at at least 5 employers, most-named first.
Python, Microsoft Excel and SQL are the skills named most so far, in 1,973, 1,145 and 900 of the 16,917 open roles whose skills we have read.
| Skill | Category | Roles | Employers |
|---|---|---|---|
| Python | Programming Languages | 1,973 | 287 |
| Microsoft Excel | Data Science, Analytics and AI/ML | 1,145 | 301 |
| SQL | Databases | 900 | 248 |
| Amazon Web Services (AWS) | Cloud Computing and DevOps | 808 | 192 |
| Kubernetes | Cloud Computing and DevOps | 717 | 156 |
| Microsoft Office | Project Management and Productivity Tools | 496 | 160 |
| Salesforce | Sales and E-Commerce | 482 | 167 |
| Data Analysis | Data Science, Analytics and AI/ML | 471 | 196 |
| Microsoft Word | Project Management and Productivity Tools | 471 | 169 |
| C++ | Programming Languages | 465 | 93 |
| Java | Programming Languages | 438 | 115 |
| Microsoft Azure | Cloud Computing and DevOps | 426 | 132 |
| Go | Programming Languages | 419 | 101 |
| TypeScript | Programming Languages | 389 | 122 |
| Linux | IT Infrastructure and Networking | 366 | 96 |
| CI/CD | Cloud Computing and DevOps | 358 | 122 |
| Machine Learning | Data Science, Analytics and AI/ML | 338 | 126 |
| Docker | Cloud Computing and DevOps | 334 | 117 |
| JavaScript | Web Development | 327 | 119 |
| Terraform | Cloud Computing and DevOps | 314 | 105 |
How many job postings ask for AI skills?
1,366 of the 16,917 open roles whose skills we have read so far ask for at least one AI skill, at 270 employers.
Machine Learning (338 roles), PyTorch (205 roles) and Artificial Intelligence (AI) (176 roles) are the AI skills asked for most.
| AI skill | Category | Roles | Employers |
|---|---|---|---|
| Machine Learning | Data Science, Analytics and AI/ML | 338 | 126 |
| PyTorch | Data Science, Analytics and AI/ML | 205 | 69 |
| Artificial Intelligence (AI) | Data Science, Analytics and AI/ML | 176 | 70 |
| Generative AI | Data Science, Analytics and AI/ML | 147 | 67 |
| LLMs (Large Language Models) | Data Science, Analytics and AI/ML | 127 | 75 |
| TensorFlow | Data Science, Analytics and AI/ML | 118 | 54 |
| Prompt Engineering | Data Science, Analytics and AI/ML | 100 | 65 |
| Retrieval-Augmented Generation (RAG) | Data Science, Analytics and AI/ML | 97 | 49 |
| LangChain | Data Science, Analytics and AI/ML | 95 | 36 |
| Reinforcement Learning | Data Science, Analytics and AI/ML | 77 | 32 |
What counts as an AI skill (193 skills)
Machine learning and deep learning and their methods; generative AI, large language models, agents and prompting; and AI models, platforms, frameworks and AI-branded tools. Analytics and BI tools (Excel, Tableau, Power BI), classical statistics and data engineering are not counted, even where the taxonomy files them under “Data Science, Analytics and AI/ML”. A role counts once however many it names. The full list:
Agent2Agent Protocol (A2A) · Agentforce · Agentic systems · Agentic Workflows · AI Agents & Agentic AI · AI Enablement · AI Integrations · AI Orchestration · AI Workflows · Artificial Intelligence · Artificial Intelligence (AI) · BERT · Chatbot · Context Engineering · Custom GPTs / GPT Builder · Embeddings · Few-Shot Learning · Fine-tuning · GenAI (Generative AI) · Generative AI · Generative AI Agents · Generative Model Architectures · Large Language Modeling · LLM Application · LLM Fine-Tuning · LLMs (Large Language Models) · Model Context Protocol · Model Context Protocol (MCP) · Multi-Agent Systems · Multimodal Prompts · Prompt Chaining · Prompt Engineering · Prompt Engineering Tools · Prompt Patterns · Responsible AI · Retrieval-Augmented Generation · Retrieval-Augmented Generation (RAG) · Token Optimization · Tool Calling · Vector Databases · Virtual Personal Assistant Development · Applied Machine Learning · Artificial Neural Networks · Autoencoders · Automated Machine Learning (AutoML) · Classification Algorithms · Computer Vision · Convolutional Neural Networks (CNN) · Data Augmentation · Decision Tree Learning · Deep Learning · Deep Reinforcement Learning · Ensemble Methods · Face Detection · Feature Engineering · Federated Learning · Generative Adversarial Networks (GAN) · Generative Adversarial Networks (GANs) · Hyperparameter Tuning · Machine Learning · Machine Learning Algorithms · Machine Learning Methods · Machine Learning Software · ML · ML Model Deployment · ML Model Evaluation · ML Model Monitoring · ML Orchestration · MLOps · MLOps (Machine Learning Operations) · Model Compression & Quantization · Model Deployment · Model Evaluation · Model Optimization · Model Training · Model Tuning · Natural Language Processing · Natural Language Processing (NLP) · Neural Networks · Object Detection · Random Forest Algorithm · Recommendation Engine · Recurrent Neural Networks (RNNs) · Reinforcement Learning · Sentiment Analysis · Statistical Machine Learning · Supervised Learning · Supervised Machine Learning · Transfer Learning · Unsupervised Learning · Unsupervised Machine Learning · Vision Transformer (ViT) · Adobe Firefly · Amazon Bedrock · Amazon Q · Amazon SageMaker · Anthropic Claude AI · Anthropic Claude API · AssemblyAI · AutoKeras · AWS Rekognition · AWS SageMaker · Azure AI Services · Azure Machine Learning · Baseten · BeeAI · BentoML · Chainer · ChatGPT · Chroma · ClearML · Core ML · CrewAI · DALL·E · Deepgram · Deeplearning4j · Deepseek · DeepSeek API · Detectron2 · Determined AI · Dialogflow · Dify · Fireworks AI · Gemini · Google Agent Development Kit (ADK) · Google Cloud Vertex AI · Google Gemini · Google Gemini (Bard) · Google NotebookLM · Google Vertex AI · Grok · H2O.ai · Hugging Face · Hugging Face Transformers · IBM Watson · JAX · Keras · Keras (Neural Network Library) · Kubeflow · LangChain · LangGraph · LlamaIndex · Microsoft Agent Framework · Microsoft Foundry · Milvus · MLflow · MMDetection · Natural Language Toolkit (NLTK) · Neptune.ai · NVIDIA Triton Inference Server · Ollama · OpenAI · OpenAI Agents SDK · OpenAI API · OpenCV · PaddlePaddle · Perplexity AI · Pinecone · PyTorch · Rasa AI Platform · Ray/Anyscale · Scikit Learn · scikit-learn · Seldon Core · Semantic Kernel · Stable Diffusion · TensorFlow · Turi Create · Weka · XGBoost · AI Art Generation · AI Content Generation · AI Ethics and Governance · AI literacy · AI Personalization · AI powered creativity · AI Product Strategy · AI Security · Anthropic Claude Code · Cursor AI · Gamma AI · Generative Engine Optimization (GEO) · GitHub Copilot · Leonardo.Ai · Lovable AI · Microsoft Copilot · Midjourney · Miro AI · OpenAI Codex · OpenAI Sora · Spring AI · Suno AI · Windsurf AI
Which kinds of skills do job postings ask for most?
A role counts once in each category it names a skill from, so the shares add to more than 100%.
Data Science, Analytics and AI/ML (3,727 roles), Programming Languages (2,606 roles) and Project Management and Productivity Tools (2,404 roles) are the skill categories named in the most postings.
How this is measured
- A crawled role’s skills come from a language model reading its posting and matching what it names to our skills taxonomy; an employer-posted role carries the skills the employer chose. Skills the taxonomy does not hold yet are not counted.
- Every share is of the roles whose skills have been read, never of all open roles. Most open roles had not been read when the backlog was measured on 2026-09-16, and the parser only reads roles whose full description we have fetched and by default skips roles known to be weeks old, so the roles read are not a random sample. Below 50% of open roles or 2,000 roles read, this page gives counts and no percentages.
- A skill is ranked only when at least 5 employers ask for it, so one employer’s many openings cannot make a skill “in demand”.
- No “fastest-rising skills” yet: we started keeping a daily record of these counts in October 2026, and while the backlog drains any rise would measure our parser catching up, not employers asking for more.
- These are the boards we track — hundreds of employers, not the whole market. Updated hourly; each skill’s own page has its employers, places, pay and the skills asked for alongside it. See hiring trends for the rest of the numbers.
Cite these numbers
Free to reuse under CC BY 4.0 with a link to this page. Please give the date and the coverage — the numbers are re-read hourly and the share of roles read is still growing. Figures as of 2026-10-28; the full table is in the CSV.
Career App, "The most in-demand skills in job postings", career.thegoodapps.co/trends/skills, accessed 2026-10-28.