# Job Skill Enhancement System

## CORE INSTRUCTION:
You are an AI that extracts and identifies coding languages and frameworks from job descriptions. Your goal is to find ALL relevant technical skills mentioned in the job posting, even if they're mentioned in different ways or aliases.

## INPUT DATA:
{
  "job": {"job_id":5040,"title":"Head of AI Personalization","company_name":"Danske Bank","description":"About the job\n\nJob Description\n\nHead of AI Personalization\n\nAbout The Role\n\nWe are looking for an experienced and visionary Team Leader to lead the newly established AI Personalization team. You will be leading a group of highly skilled data scientists who are ready to redefine how customers experience banking through data-driven insights and personalization. You will be responsible for both the strategic direction and day-to-day delivery, playing a central role in building AI solutions that create real business impact and an improved customer experience.\n\nWho Are We\n\nThe AI Personalization team sits in the Personal Customers AI CoE - a diverse team of Data Scientists, ML Engineers with various educational and cultural backgrounds. We are responsible for applying AI in Danske Bank’s customer facing solutions, ranging anywhere from ML models used for personalisation and marketing optimisation to chatbots and advisor assistants.\n\nWe pride ourselves on being at the forefront of the latest technology and research, identifying use cases where AI can be applied to create value for our customers and the bank.\n\n💼 What You’ll Do\n\n\nLead, motivate, and develop a team of data scientists (4 full time and 1 student), ensuring high-quality delivery and continuous professional growth\nDefine roadmaps and priorities together with stakeholders (e.g., product owners, business leads) to focus on the most value-creating projects\nContribute to modelling, data transformation, feature engineering, validation, and deployment of advanced statistical and machine learning models\nEnsure best practices for data pipelines, code conventions, version control, and model governance\nOversee model monitoring, performance evaluation, and operational stability in production\nSupport experiment and A\/B test design, result evaluation, and clear communication of insights to both technical and non-technical stakeholders\nDrive innovation by evaluating and adopting new tools, technologies, and methods within machine learning, AI, and data engineering\n\n\n📊 What You Bring\n\n\nSeveral years of experience as a data scientist, with at least 1–2 years of leadership or technical lead responsibilities for other data scientists\nStrong expertise in statistical methods and machine learning, including models such as regression, tree-based methods, neural networks, etc.\nSolid programming skills (e.g., Python, R), database experience (SQL\/NoSQL), and knowledge of MLOps practices, model deployment, and operations in production\nExperience with cloud platforms (e.g., Azure, AWS, GCP) and modern data architectures\nAbility to communicate complex technical topics clearly to non-technical colleagues and senior stakeholders\nProactive, structured, and quality-oriented; able to work strategically while staying close to the daily deliveries\nYou must hold a Danish Working Permit or live in Denmark to be eligible for the role.\n\n\n🌟 Nice to Have\n\n\nExperience with marketing automation or personalization platforms (e.g., Salesforce, Adobe, Pega)\nFamiliarity with uplift modelling, causal ML, or reinforcement learning for offer optimization\nExperience deploying models into production using MLOps tools (e.g., MLflow, Airflow, Docker, Azure ML)\n\n\n🚀 Why Join Us?\n\n\nHelp shape how the bank interacts with its customers using advanced analytics and AI\nJoin a growing, ambitious AI team working at the core of our digital transformation\nCollaborate across an agile and business-oriented organization\nWorking in our new HQ with great facilities (2-3 days per week)\n\n\n📩 Interested?\n\nPlease send your CV and a short motivational letter including:\n\n\nKey data science projects you have led and their business impact\nYour leadership or technical lead experience\nThe technologies and methods you are most confident with and those you wish to further develop\n\n\nApply now or reach out to Rasmus Gansted Brink – RBRI@danskebank.dk for more information. Let’s build the future of AI together.\n\nDeadline for applications is Wednesday 8th October.","brief_summary_of_job":null,"existing_skills_from_job":["Python","NoSQL","AWS","Azure","GCP"],"existing_skills_from_database":[]},
  "coding_categories": {"languages":["Bash","C","C#","C++","CSS","Clojure","Dart","Elixir","Go","Groovy","Haskell","HTML","Java","JavaScript","Julia","Kotlin","Lua","MATLAB","Objective-C","Perl","PHP","PowerShell","Python","R","Ruby","Rust","Sass","Scala","Swift","TypeScript"],"frameworks":[".NET","Angular","Apache Spark","ASP.NET","Backbone.js","Bootstrap","CodeIgniter","Django","Ember.js","Express","FastAPI","Flask","Flutter","Hadoop","Ionic","jQuery","Laravel","Livewire","Meteor","NestJS","Next.js","Node.js","Nuxt.js","Phoenix","PHPUnit","PyTorch","React","React Native","Ruby on Rails","Spring Boot","Svelte","Symfony","Tailwind CSS","TensorFlow","Vue.js","Xamarin","Alpine.js","Filament","WordPress"],"databases":["MySQL","PostgreSQL","Redis","MongoDB","DynamoDB","MariaDB","NoSQL","Oracle","BigQuery","Elasticsearch","SQL","SQL Server","SQLite","Cassandra","CouchDB","Neo4j","InfluxDB","CockroachDB"],"tools":["Git","GitHub","GitLab","Bitbucket","Docker","Kubernetes","CI\/CD","Jenkins","Kafka","RabbitMQ","Amazon SQS","AMQP","PubSub","REST API","RESTful APIs","GraphQL","AWS","Azure","GCP","Terraform","Ansible","Puppet","Chef","Vagrant","Vault","Consul","Prometheus","Grafana","ELK Stack","Splunk"],"skills":["English","Danish","Communication","Full-stack development","Back-end development","Front-end development","Cloud computing","DevOps","Microsoft Excel","PowerBI","Power Platform","Agile","Scrum","Problem-solving","Team collaboration","Physical presence","Remote work","Hybrid work"],"colors":{"languages":"blue","frameworks":"purple","databases":"orange","tools":"indigo"}},
  "skill_aliases": {"bash":"Bash","c":"C","c plus plus":"C++","C plus plus":"C++","c sharp":"C#","C sharp":"C#","c#":"C#","c++":"C++","clojure":"Clojure","cpp":"C++","dart":"Dart","elixir":"Elixir","go":"Go","golang":"Go","Golang":"Go","groovy":"Groovy","haskell":"Haskell","java":"Java","java script":"JavaScript","Java script":"JavaScript","javascript":"JavaScript","Javascript":"JavaScript","js":"JavaScript","JS":"JavaScript","es5":"JavaScript","ES5":"JavaScript","es6":"JavaScript","ES6":"JavaScript","julia":"Julia","kotlin":"Kotlin","lua":"Lua","matlab":"MATLAB","Matlab":"MATLAB","objective c":"Objective-C","Objective C":"Objective-C","objective-c":"Objective-C","objc":"Objective-C","ObjC":"Objective-C","obj-c":"Objective-C","Obj-C":"Objective-C","perl":"Perl","php":"PHP","Php":"PHP","powershell":"PowerShell","Power Shell":"PowerShell","python":"Python","py":"Python","Py":"Python","phyton":"Python","r":"R","ruby":"Ruby","rust":"Rust","scala":"Scala","swift":"Swift","typescript":"TypeScript","Typescript":"TypeScript","ts":"TypeScript","TS":"TypeScript","mysql":"MySQL","postgresql":"PostgreSQL","postgres":"PostgreSQL","mongodb":"MongoDB","mongo":"MongoDB","redis":"Redis","sqlite":"SQLite","oracle":"Oracle","mssql":"SQL Server","sql server":"SQL Server","dynamodb":"DynamoDB","dynamo db":"DynamoDB","mariadb":"MariaDB","maria db":"MariaDB","nosql":"NoSQL","no sql":"NoSQL","bigquery":"BigQuery","big query":"BigQuery","elasticsearch":"Elasticsearch","elastic search":"Elasticsearch","cassandra":"Cassandra","couchdb":"CouchDB","couch db":"CouchDB","neo4j":"Neo4j","neo 4j":"Neo4j","influxdb":"InfluxDB","influx db":"InfluxDB","cockroachdb":"CockroachDB","cockroach db":"CockroachDB","aws":"AWS","amazon web services":"AWS","azure":"Azure","microsoft azure":"Azure","gcp":"GCP","google cloud platform":"GCP","google cloud":"GCP",".Net":".NET","angular":"Angular","angular js":"Angular","Angular js":"Angular","angularjs":"Angular","AngularJS":"Angular","apache spark":"Apache Spark","asp net":".NET","ASP net":".NET","asp.net":"ASP.NET","ASP.NET":".NET","asp.net core":".NET","ASP.NET Core":".NET","backbone":"Backbone.js","BackboneJS":"Backbone.js","backbone js":"Backbone.js","backbonejs":"Backbone.js","bootstrap":"Bootstrap","Bootstrap framework":"Bootstrap","twitter bootstrap":"Bootstrap","Twitter Bootstrap":"Bootstrap","code igniter":"CodeIgniter","codeigniter":"CodeIgniter","django":"Django","ember":"Ember.js","ember js":"Ember.js","emberjs":"Ember.js","EmberJS":"Ember.js","express":"Express","express.js":"Express","Express.js":"Express","expressjs":"Express","ExpressJS":"Express","fast api":"FastAPI","Fast api":"FastAPI","fastapi":"FastAPI","flask":"Flask","flutter":"Flutter","hadoop":"Hadoop","Hadoop":"Hadoop","ionic":"Ionic","Ionic framework":"Ionic","ionic framework":"Ionic","jquery":"jQuery","JQuery":"jQuery","JQUERY":"jQuery","laravel":"Laravel","meteor":"Meteor","meteor js":"Meteor","meteorjs":"Meteor","MeteorJS":"Meteor","nestjs":"NestJS","nest js":"NestJS","Nest JS":"NestJS","Nestjs":"NestJS","next js":"Next.js","Next js":"Next.js","next.js":"Next.js","nextjs":"Next.js","NextJS":"Next.js","node":"Node.js","Node":"Node.js","node.js":"Node.js","nodejs":"Node.js","NodeJS":"Node.js","nuxt js":"Nuxt.js","Nuxt js":"Nuxt.js","nuxt.js":"Nuxt.js","nuxtjs":"Nuxt.js","NuxtJS":"Nuxt.js","phoenix":"Phoenix","Phoenix framework":"Phoenix","pytorch":"PyTorch","Pytorch":"PyTorch","torch":"PyTorch","Torch":"PyTorch","react":"React","react.js":"React","React.js":"React","reactjs":"React","ReactJS":"React","react native":"React Native","ReactNative":"React Native","reactnative":"React Native","rn":"React Native","RN":"React Native","rails":"Ruby on Rails","Rails":"Ruby on Rails","ror":"Ruby on Rails","ROR":"Ruby on Rails","ruby on rails":"Ruby on Rails","Ruby On Rails":"Ruby on Rails","spark":"Apache Spark","Spark":"Apache Spark","spring":"Spring Boot","Spring":"Spring Boot","spring boot":"Spring Boot","Spring Boot":"Spring Boot","spring framework":"Spring Boot","Spring Framework":"Spring Boot","SpringBoot":"Spring Boot","springboot":"Spring Boot","svelte":"Svelte","symfony":"Symfony","tailwind":"Tailwind CSS","Tailwind":"Tailwind CSS","tailwind css":"Tailwind CSS","Tailwind Css":"Tailwind CSS","tailwindcss":"Tailwind CSS","TailwindCSS":"Tailwind CSS","tensorflow":"TensorFlow","Tensorflow":"TensorFlow","tensor flow":"TensorFlow","Vue":"Vue.js","vue":"Vue.js","vue js":"Vue.js","vue.js":"Vue.js","vuejs":"Vue.js","VueJS":"Vue.js","xamarin":"Xamarin","Xamarin Forms":"Xamarin","Xamarin.forms":"Xamarin","livewire":"Livewire","phpunit":"PHPUnit","alpine":"Alpine.js","alpine js":"Alpine.js","alpinejs":"Alpine.js","filament":"Filament","docker":"Docker","kubernetes":"Kubernetes","k8s":"Kubernetes","github":"GitHub","gitlab":"GitLab","bitbucket":"Bitbucket","terraform":"Terraform","puppet":"Puppet","chef":"Chef","git":"Git","jenkins":"Jenkins","kafka":"Kafka","rabbitmq":"RabbitMQ","amazon sqs":"Amazon SQS","amqp":"AMQP","pubsub":"PubSub","rest api":"REST API","restful api":"RESTful APIs","restful apis":"RESTful APIs","graphql":"GraphQL","ansible":"Ansible","vagrant":"Vagrant","vault":"Vault","consul":"Consul","prometheus":"Prometheus","grafana":"Grafana","elk stack":"ELK Stack","splunk":"Splunk","scrum":"Scrum","agile":"Agile","kanban":"Kanban","devops":"DevOps","ci\/cd":"CI\/CD","tdd":"TDD","bdd":"BDD","linux":"Linux","windows":"Windows","macos":"macOS","ubuntu":"Ubuntu","centos":"CentOS","debian":"Debian","junit":"JUnit","pytest":"PyTest","jest":"Jest","mocha":"Mocha","cypress":"Cypress","selenium":"Selenium","html":"HTML","HTML":"HTML","html5":"HTML","HTML5":"HTML","css":"CSS","CSS":"CSS","css3":"CSS","CSS3":"CSS","sass":"Sass","scss":"Sass","SCSS":"Sass","less":"Less","LESS":"Less","webpack":"Webpack","gulp":"Gulp","npm":"npm","yarn":"Yarn","sql":"SQL","full time":"Fuldtid","full-time":"Fuldtid","fuldtid":"Fuldtid","part time":"Deltid","part-time":"Deltid","deltid":"Deltid","contract":"Kontrakt","kontrakt":"Kontrakt","remote":"Fjernarbejde","fjernarbejde":"Fjernarbejde","hybrid":"Hybridarbejde","hybridarbejde":"Hybridarbejde","on-site":"Fysisk tilstedeværelse","onsite":"Fysisk tilstedeværelse","fysisk tilstedeværelse":"Fysisk tilstedeværelse"}
}

## TASK:
1. **Extract ALL coding languages and frameworks** mentioned in the job description
2. **Use the skill_aliases mapping** to normalize skill names (e.g., "JS" → "JavaScript", "React.js" → "React")
3. **Only include skills from the coding_categories** (languages and frameworks)
4. **Avoid duplicates** - if "JavaScript" and "JS" are both mentioned, only include "JavaScript"
5. **Be thorough** - look for skills mentioned in:
   - Job title
   - Job description
   - Brief summary
   - Requirements sections
   - Nice-to-have sections

## SKILL CATEGORIES TO EXTRACT:
- **Languages**: Programming languages (PHP, Python, JavaScript, Java, C#, Go, Rust, etc.)
- **Frameworks**: Web frameworks and libraries (Laravel, React, Vue.js, Angular, Django, Spring Boot, etc.)
- **Databases**: Database systems and data stores (MySQL, PostgreSQL, Redis, MongoDB, DynamoDB, etc.)
- **Tools**: Development tools and infrastructure (Git, Docker, AWS, CI/CD, Jenkins, Kafka, etc.)

## EXCLUSION RULES:
- Do NOT extract general skills (communication, languages like English, etc.)
- Do NOT extract business skills (Excel, PowerBI for non-technical roles)
- Do NOT extract work arrangements (remote, physical presence)
- Do NOT include methodologies (Agile, Scrum, etc.) unless specifically technical
- Do NOT include skills already in existing_skills_from_database

## OUTPUT FORMAT:
Return ONLY valid JSON in this exact structure:
{
  "job_id": <integer>,
  "extracted_skills": [
    {
      "skill_name": "<normalized_skill_name>",
      "category": "<languages|frameworks|databases|tools>",
      "confidence": <integer 0-100>,
      "mentioned_as": ["<original_mention_1>", "<original_mention_2>"],
      "context": "<brief_context_where_found>"
    }
  ],
  "reasoning": {
    "total_skills_found": <integer>,
    "skills_by_category": {
      "languages": ["<skill1>", "<skill2>"],
      "frameworks": ["<skill1>", "<skill2>"],
      "databases": ["<skill1>", "<skill2>"],
      "tools": ["<skill1>", "<skill2>"]
    },
    "duplicates_removed": ["<original_mention> → <normalized_name>"],
    "excluded_skills": ["<skill>", "<reason>"]
  },
  "confidence": <integer 0-100>
}

Return nothing outside the JSON object.

{
  "job_id": 5040,
  "extracted_skills": [
    {
      "skill_name": "Python",
      "category": "languages",
      "confidence": 100,
      "mentioned_as": ["Python"],
      "context": "Solid programming skills"
    },
    {
      "skill_name": "R",
      "category": "languages",
      "confidence": 100,
      "mentioned_as": ["R"],
      "context": "Strong expertise in statistical methods and machine learning"
    },
    {
      "skill_name": "SQL",
      "category": "databases",
      "confidence": 100,
      "mentioned_as": ["SQL"],
      "context": "Database experience"
    },
    {
      "skill_name": "NoSQL",
      "category": "databases",
      "confidence": 100,
      "mentioned_as": ["NoSQL"],
      "context": "Database experience"
    },
    {
      "skill_name": "AWS",
      "category": "tools",
      "confidence": 100,
      "mentioned_as": ["AWS"],
      "context": "Experience with cloud platforms"
    },
    {
      "skill_name": "Azure",
      "category": "tools",
      "confidence": 100,
      "mentioned_as": ["Azure"],
      "context": "Experience with cloud platforms"
    },
    {
      "skill_name": "GCP",
      "category": "tools",
      "confidence": 100,
      "mentioned_as": ["GCP"],
      "context": "Experience with cloud platforms"
    },
    {
      "skill_name": "Docker",
      "category": "tools",
      "confidence": 100,
      "mentioned_as": ["Docker"],
      "context": "Experience deploying models into production using MLOps tools"
    },
    {
      "skill_name": "MLflow",
      "category": "tools",
      "confidence": 100,
      "mentioned_as": ["MLflow"],
      "context": "Experience deploying models into production using MLOps tools"
    },
    {
      "skill_name": "Airflow",
      "category": "tools",
      "confidence": 100,
      "mentioned_as": ["Airflow"],
      "context": "Experience deploying models into production using MLOps tools"
    }
  ],
  "reasoning": {
    "total_skills_found": 10,
    "skills_by_category": {
      "languages": ["Python", "R"],
      "frameworks": [],
      "databases": ["SQL", "NoSQL"],
      "tools": ["AWS", "Azure", "GCP", "Docker", "MLflow", "Airflow"]
    },
    "duplicates_removed": [],
    "excluded_skills": []
  },
  "confidence": 100
}