Simon Willison
Co-creator of Django; writes a widely cited independent LLM blog
Co-created the Django web framework and now documents large language models in public at simonwillison.net, including the coining of the term "prompt injection". He also builds the open-source LLM command-line tool.
Key arguments & positions
- Coined "prompt injection" in 2022 and argues it remains structurally unsolved, so any agent handling untrusted text should be designed around that.
- Argues the useful way to judge a model is to use it daily on real work and write down what happened, rather than to read benchmark tables.
- Advocates local and open-weight models for privacy-sensitive work, and documents how to run them.
Accomplishments
- Co-created the Django web framework (2005) and co-founded Lanyrd.
- Creator of Datasette and of the open-source llm command-line tool.
- Publishes one of the most-cited independent running records of LLM behavior at simonwillison.net.
Papers & key writings
Links
In the library
Nothing of theirs is filed in the hub yet. Browse the full library.
Recommended next
Hand-picked from the hub based on what Simon Willison covers.
garak
An open-source scanner that probes a language model for weaknesses — prompt injection, data leakage, jailbreaks, toxic output — and reports what it found. Free.
Why this: Covers prompt injection and open source too
Reactive Resume (rxresu.me)
This is a free resume builder with no paywalls, ads, or watermarks. You can choose from multiple templates to create, edit, and export ATS-friendly resumes with custom sections.
Why this: Also about open source
Job Search Agent (open-source AI job tracker)
This open-source tool scans job boards daily to find openings that fit your criteria. You can connect your LinkedIn network to spot contacts at hiring companies and receive daily email updates with active job matches.
Why this: Also about open source
vLLM
An open-source engine for serving large language models with high throughput and efficient memory use. Its PagedAttention approach helps reduce wasted GPU memory during inference.
Why this: Also about open source