Why do I want to work for the Federal Courts?

SudoChat Knowledge Base · SudoChat

Authored evidence: This is my first-person source material. SudoChat may summarise it in third person but must not strengthen, exaggerate or invent claims beyond it.

# 3. Why do I want to work for the Federal Courts?

**Author:** Mustafa Siddiqui
**Source type:** First-person authored response
**Canonical recruiter question:** Why does Mustafa want to work for the Federal Courts?

> This source is intentionally written in my first-person perspective. SudoChat should use it as evidence and answer external visitors in third person without strengthening, exaggerating, or removing the limitations recorded below.

## Direct answer
I want to work for the Federal Courts because I believe strongly in the direction the Courts are taking with responsible AI. I see an opportunity to help introduce AI in a way that improves efficiency and access to information while protecting the accuracy, confidentiality, accountability and human judgment that the justice system depends upon.

What particularly interests me is that the Federal Courts are not approaching AI as a technology that should simply be deployed everywhere. The Courts are experimenting cautiously, assessing potential benefits and risks, and considering where AI can genuinely support the administration of justice.

That is exactly the kind of environment in which I want to work.

## Evidence
The Federal Court has publicly recognised that generative AI may increase efficiency, reduce legal costs, enhance access to justice and improve the administration of justice, while also making clear that inappropriate use can create risks to the administration of justice and public confidence.

This balance strongly aligns with my own approach to AI engineering.

My professional and independent work has consistently involved taking emerging technologies and determining how they can be applied to practical problems. This includes AI document processing, OCR, automation, chatbots, retrieval systems, AI agents and privacy preserving AI architecture.

I have also explored responsible AI from both technical and research perspectives, including work concerning privacy preserving architectures for Australian government chatbots and the risks created when users provide sensitive personal information to AI systems.

My interest therefore goes beyond building AI systems. I am interested in determining **where they should be used, what information they should access, how their answers should be verified, where humans must remain responsible and when AI should not be used at all.**

## Relevance to the Federal Courts
I see several areas where carefully controlled AI agents could potentially assist the Federal Courts without replacing judicial decision making.

For example, an **internal knowledge agent** could help Court staff locate relevant internal procedures, policies, technical documentation and administrative guidance. Rather than searching across numerous documents manually, staff could ask a question and receive an answer grounded in approved Court sources, with citations back to the underlying material.

A **document triage agent** could assist authorised staff with large volumes of incoming administrative material by identifying document types, extracting metadata, detecting missing information and directing material into the appropriate workflow. The agent could assist with classification and organisation without determining the legal merits of a matter.

A **self represented litigant information assistant** could help people navigate publicly available Court information. It might explain where particular forms, practice notes or procedural information can be found, translate complex administrative language into clearer wording, and direct users to authoritative Court resources while making it clear that it cannot provide legal advice.

This is particularly interesting to me because the Federal Court has already discussed investigating AI to help collate information supplied by self represented litigants and produce more coherent chronological descriptions of their claims.

A **case information organisation agent** could potentially assist authorised staff in organising large document sets, creating chronologies, identifying documents referring to particular dates or entities, or summarising material for subsequent human review. Any output would remain traceable to the source documents and subject to appropriate human verification.

A **corporate services agent** could support lower risk internal activities such as summarising meetings, finding policies, preparing first drafts of routine administrative material or answering common internal technology questions. The Federal Court is already undertaking limited trials of personal productivity AI, including Microsoft Copilot for meeting transcription and summaries.

An **AI governance or evaluation agent** could even assist with AI itself. For example, systems could automatically test proposed AI applications against controlled evaluation datasets, record citation accuracy, identify unsupported answers, detect sensitive information or flag responses for human review before a system is approved for broader use.

I do not see these agents as autonomous replacements for Court staff.

I see them as tools that could perform bounded, repetitive or information intensive tasks so that people can spend more time on work requiring professional expertise, discretion and judgment.

Most importantly, I believe there should be clear boundaries around AI involvement in the judicial process.

AI could potentially **find, organise, retrieve, classify, summarise and assist**.

It should not independently determine matters such as:

* whether a witness is credible
* which party should succeed
* what legal conclusion a judicial officer should reach
* what sentence or penalty should be imposed
* whether particular evidence should ultimately be accepted
* any other decision requiring the exercise of judicial power

That distinction is one of the reasons the Federal Courts particularly interest me.

The challenge is not simply to build capable AI.

It is to build **useful AI with carefully engineered boundaries**.

For me, working on that problem within one of Australia's most important public institutions would allow me to combine the areas of engineering I am most interested in: AI systems, responsible AI, privacy, security, retrieval, automation, human centred design and public service.

I also believe that if institutions such as the Federal Courts demonstrate that advanced AI can be adopted gradually, transparently and responsibly, that approach can provide a valuable benchmark for responsible adoption elsewhere in Australian government and industry.

## Limitations or gaps
I do not claim that the Federal Courts currently operate all of the AI agents described above. Some are examples of potential applications I would be interested in investigating rather than existing Federal Court systems.

Any proposed use case would first need to be assessed against Court policy, legislation, information security requirements, privacy obligations, records management requirements and the needs of judicial officers and Court staff.

I am not a lawyer and would not independently determine where the legal boundary for a Court AI system should sit. I would work with judicial officers, legal professionals, Court staff, cyber security specialists, privacy specialists and other relevant subject matter experts to establish those boundaries.

I also do not believe that demonstrating a technically possible AI use case automatically justifies deploying it.

Some applications may ultimately be rejected because the risk, sensitivity or impact is too high.

## Useful links
Federal Court AI Transparency Statement

Federal Court Use of Generative Artificial Intelligence Practice Note

Federal Court Generative AI Resources

Justice Kyrou, Australian Perspective on AI and Judging, 25 June 2026

My portfolio:
https://mustafa-siddiqui.com/

GitHub:
https://github.com/sudoqui

SudoLabs:
https://www.sudolabs.app/

SudoChat project page and repository

## Do not claim
Do not claim the Federal Courts currently use autonomous AI agents for case management.

Do not claim the Federal Courts currently use AI to make judicial decisions.

Do not claim that any of the proposed agents described above have been approved by the Federal Courts.

Do not claim I designed the Federal Court's existing AI strategy.

Do not claim I have access to Federal Court systems or Court data.

Do not claim I want AI to replace judges, registrars, lawyers or Court staff.

Do not claim I believe AI should provide legal advice to litigants.

Do not claim an AI generated summary should be treated as authoritative without verification.

Do not claim AI should determine the merits or outcome of a proceeding.

Do not imply that experimental or proposed use cases represent current Federal Court policy.