Artificial intelligence and automation
Systems that learn from data, generate language, or act in the physical world, and the engineering needed to make them dependable.
Artificial intelligence
- What it means
- Building systems that perform tasks normally requiring human judgement, such as understanding language, recognising patterns and planning.
- Problem it solves
- Work that is too large, too fast or too repetitive for people to do by hand.
- Practical applications
- Document and language processing
- Decision support
- Search and recommendation
- Technologies and methods
- Python
- Language-model APIs
- Search and retrieval
- Evaluation methods
- Potential projects
- An assistant that answers questions from an organization’s own documents
- A triage tool that sorts incoming requests
Machine learning
- What it means
- Training statistical models on data so that they can predict, classify or detect, and measuring honestly how well they do it.
- Problem it solves
- Questions where the rules are too complex to write down but examples are plentiful.
- Practical applications
- Classification and forecasting
- Anomaly detection
- Model evaluation and monitoring
- Technologies and methods
- scikit-learn
- PyTorch
- pandas
- Jupyter
- Cross-validation
- Potential projects
- A classifier with a documented evaluation
- A demand forecast with stated error bounds
Generative AI
- What it means
- Using large language and media models to produce text, code or images, with controls that keep the output grounded and safe.
- Problem it solves
- Slow drafting, summarising and question-answering over large bodies of text.
- Practical applications
- Retrieval-grounded assistants
- Drafting and summarisation tools
- Structured extraction from documents
- Technologies and methods
- Large language models
- Retrieval-augmented generation
- Prompt design
- Output evaluation
- Potential projects
- A grounded website assistant, such as the one on this site
- A summarisation tool for long reports
Robotics and automation
- What it means
- Programming machines and processes to sense, decide and act with little human intervention.
- Problem it solves
- Manual processes that are slow, error-prone or unsafe for people.
- Practical applications
- Process automation
- Sensor-driven control
- Prototyping with microcontrollers
- Technologies and methods
- Python
- Microcontrollers
- Sensors and actuators
- Control loops
- ROS
- Potential projects
- A sensor-based monitoring prototype
- An automated workflow that replaces a manual office process
Related services
Have a project or a question?
Describe what you need. Kansi reads every enquiry and replies by email.