How to become an AI Engineer

Background

My #universityofoxford course on #AI is almost full. Uniquely, this year, the course is oriented towards being an #AIengineer.?

This term, AI Engineer, is?a complex term. One well known book defines an AI Engineer as developing using LLMs. In my view, this definition is limiting.

I define an AI engineer more broadly covering three aspects:

1. Developing AI systems with the help of LLMs(prompt engineering, RAG and fine tuning)
2. LLM assisted code development (Claude, OpenAI, Github copilot etc)
3. Machine learning and deep learning based systems that are not based on LLMs??

This is a complex but pragmatic definition - especially for people who want to transition their career to AI.? Based on my work in one of the largest rollouts of AI, the complexity is actually based on the business processes and the data?

Thus, an AI product is more than just a model; it combines AI technology with software engineering, design, and business processes to deliver a practical solution that solves real-world problems. We also need to consider data and people. One good example is the GSK approach ?by Shobie Ramakrishnan

Lets break down? the idea of an AI Engineer

What is an AI product

An AI product is a software application or system that leverages artificial intelligence (AI) techniques to solve problems, improve processes, or enhance user experiences. It typically integrates AI models, algorithms, or systems into a functional solution that delivers value to end-users.? ?

Components of an AI Product

1. AI Models: Algorithms trained on data to perform specific tasks.
2. Data Infrastructure: Systems to collect, process, and store data.
3. User Interface (UI): Frontend through which users interact with the product.
4. API or Middleware: Connects the AI component with the application layer.
5. Deployment Pipeline: Ensures efficient rollout and updates of the AI model.

Skills

We can then think of an AI Engineer in terms of skills

• Machine learning and deep learning - using Python (without LLM code generation)
• Data and pipelines
• Cloud
• Software engineering
• Model evaluation
• MLOps
• Model evaluation
• Prompt engineering, /rag and fine tuning
• Open source AI models
• Using LLM APis like OpenAI api
• Autonomous AI Agents
• AI evaluation metrics (RAG, AI as a Judge etc)
• Graphrag and causal?
• RAI and Ethics
• Small language models?
• The roadmap to AGI
• Reasoning and AI

You also need to understand the basics of model deployment ex using docker and terraform alongwith azure container services, Amazon ECS Elastic container services or GCP cloudrun

Have I missed anything??

The most important skill .. The AI Engineering Mindset

I shared about a book I read called developing the engineering mindset? by Guru Madhavan .

Any engineering (not just AI engineering) is a tradeoff between constraints, resources, performance, risk, stakeholder alignment and also includes design, aesthetics, regulation etc.

With that in mind, I can actually reframe this conversation by thinking in another way

You need to know all the options (like a palette) and the tradeoffs using those options

You need to understand the known unknowns?

Finally, you need to understand where you are (personally in your own journey) to navigate where you want to go.

This is why my teaching is reflective / Oxford style?

Finally ..

If you want to study #AI with me at the #universityofoxford - please see my course on AI (almost full now - only last few places remaining) https://conted.ox.ac.uk/courses/artificial-intelligence-generative-ai-cloud-and-mlops-online

If you want to be a part of my community see Creating a community (LinkedIn group) for my blog - where you can ask me questions re AI

If you want to work with me, we are recruiting