Data Science Career VS Academia

Differences in Data Science in regards to job market vs academia:

  • Academic vs Pragmatic:
    • Academia: creates researches with rigid expectation of environment(data, stakeholders, etc)
    • Business: requires pragmatic data scientists, with high adaptation in real life scenarios
    • Solution: Understand stakeholders, goals, values and success metrics of business and work for and toward them.
  • Acceptable performance
    • Academia: Optimizing to beat previous record
    • Business: solution is useless, even dangerous if works poorly in rare but important examples
  • Meaning of the right solution
    • Academia: better benchmark results
    • Business: Value generation
    • Solution: Understand business needs, expectation and practice proper communication.
  • Academia cares for better algorithms and techniques, Companies for better business solutions
    • Academia: Optimization, improvement, innovation
    • Business: time and money cost, interoperability, application
    • Solution: practice data-centric automation(in data handling, merging, validation and evaluation), data model interoperability and flexibility, result oriented analysis
  • Constraints
    • Academia: research with predefined constraints
    • business: performing applied research with constantly changing conditions
    • Solution: Create a model for a specific goal and practice adopting to changes extending and maintaining your model from different sources of data one by one.
  • Optimization costs
    • Academia: Everything that can be optimized should be
    • Business: optimization must be worth the cost
    • Solution: Work with different deadlines and computation resources. Practice optimization of work process.
  • Data types:
    • Academia: Static and often large datasets, which are already made and ready to use.
    • Business: You have to collect and prepare data. also dynamic often inconsistent data streams from different sources, qualities, and volume
    • Solution: Practice with data from social media sites, and other highly dynamic sites like online shops, job boards, etc.
  • Data structure:
    • Academia: specific features exists. data cleaning and feature extraction create a clean dataset.
    • Business: New features are added, data is missing, volume over time varies.
    • Solution: Practice merging data sets with varying features, data formats, thresholds, etc
  • Process of working with data:
    • Academia: Train model → validate, optimize → publish
    • business: create and deploy model, then continuously import new data and maintain model
    • Solution: practice with dynamic data from APIs or scrapped from the web
  • Stakeholders and supervisors:
    • Academia: You work with academics, with same expertise. they understand the goal, process, and analytical result
    • Business: You have to generate meaning from data in a none technical manner, for audience with differing levels of knowledge.
    • Solution: Master data visualization and presentation techniques
  • Development environment and technology
    • Academia: Standard code, libraries, structured data, defined process, specific benchmark and metrics
    • Business: Different development environments, older or domain specific software and tools, application specific output
    • Solution: Practice Software Engineering, DevOps(and MLOps), and workflows and their industry standard tools
  • Concerns:
    • Doesn’t concern itself with drift, continuous development, flexibility toward new data and features.
    • Continuous monitoring is essential, data drift poses a critical risk to businesses.