Tim Thacker
CTO and Co-Founder at Nullify
I've been programming since I was 15 and since then have been passionate about building software systems and seeing users get value out of them. At university I led the effort to build out a team of software engineers as part of the UNSW motorsport team, Redback Racing. After studying software engineering I worked for an access control security startup and then a cyber security engineering company where I helped build many products for enterprise security. I also taught the AWS cyber security course at UNSW during this time which led into starting nullify.ai with my colleague Shan Kulkarni.
In the last few years I have been a heavy user of AWS serverless, mainly Lambda and Fargate for compute. Although I started my studies doing a lot of electrical engineering and low level software, I like solving problems for my users so I let the cloud provider handle as much as possible.
At Nullify we have been building AI agents for application security. Nullify integrates with tools software engineers use (GitHub, Slack etc.) to perform tasks such as:
- detect, prioritise and fix security vulnerabilities in your codebase
- generate incident response plans and execute them
- educate engineers about secure coding practices
In this talk I will go deep into the design decisions we made when using serverless architecture to build this product which is used by some of Australia's largest enterprise organisations.
Some notable points of discussion include:
- when we use AWS Lambda or Fargate
- designing the event driven architecture using EventBridge, SNS and SQS
- the data architecture to give the agents relevant memory and knowledge
- the RAG system to provide the LLMs with the data they need to perform meaningful work
