Discover how a past winning project turned advanced AI work into clear national benefit. This introduction shows the structure and choices that framed an award‑winning case in AI and cybersecurity, and how to map a persuasive research story for peer review.
The scheme’s assessment splits focus across investigator capability, project quality, benefit and feasibility. Funding included salary support of $112,897 (30% on‑costs) and up to $50,000 per year in project support, both for three years. Weightings: Investigator/Capability 35%, Project Quality and Innovation 35%, Benefit 15%, Feasibility 15%.
Use this decra computer science template to shape a project that moves machine learning research into application‑ready milestones. Anchor evidence in ARC language, position your phd achievements as leadership signals, and show industry partnerships that strengthen translation and impact.
We’ll highlight how to create new knowledge with a pragmatic delivery plan over a three‑year horizon and point to practical opportunities for partnership and measurable outcomes. Finish ready with a short checklist that aligns learning goals with scheme expectations.
Key Takeaways
- Map investigator strengths and past phd achievements to the 35% capability criterion.
- Frame project quality with clear milestones, risk controls and feasibility over three years.
- Translate machine learning innovation into realistic applications and industry pathways.
- Quantify national benefit and post‑award delivery to meet the 15% impact weighting.
- Use concise, ARC‑aligned language to make novel work assessable and compelling.
Why this guide matters for early career researchers in Australia
This guide maps clear steps so early career researchers can turn an idea into a competitive, fundable project. It focuses on what assessors value: excellence, leadership and measurable national benefit.
Read it to learn how to scope a three‑year project, show who benefits, and craft impact that the Australian Government recognises.
User intent and outcomes: turning a template into a winning DECRA
Practical outcomes: applicants gain a roadmap to design milestones, risk controls and partner roles that make research deliverable within a year‑by‑year plan.
- How to describe who benefits and when, so reviewers see clear public value.
- Ways to show emerging leadership through supervision and collaboration.
- Scoping advice to balance ambition and feasibility while maximising impact.
- Opportunities to build national and international links that lift translation.
Use this section to shape a convincing story: why you, why now, and why this project matters for Australia’s security, economy and research ecosystem.
Understanding the ARC DECRA scheme and Discovery Program context
Here we outline what the scheme provides and how to align a research plan with the Discovery Program’s priorities.
Objectives: excellence, leadership, collaboration and national benefit
The program supports outstanding early‑career researchers with capacity for high‑quality research and emerging leadership.
Collaboration matters: projects should show links to national or international partners that strengthen methods, data access and translation.
“The scheme seeks excellence research that creates new knowledge and delivers benefit to Australia.”
Funding snapshot: salary, project support and duration
Funding: salary of $112,897 (including 30% on‑costs) for three consecutive years and up to $50,000 per year in project support.
Applicants can propose part‑time arrangements extending the award to six years where relevant to career interruptions.
- Map objectives to your project, showing systems for delivery across the grant years.
- Prioritise Investigator/Capability and Project Quality (each 35%), then Benefit (15%) and Feasibility (15%).
- Show clear development milestones that connect design, prototyping, evaluation and dissemination.
| Item | Amount / Weighting | Purpose |
|---|---|---|
| Salary | $112,897 per year | Support investigator time and leadership development |
| Project funding | Up to $50,000 per year | Equipment, data, travel and research systems |
| Assessment criteria | 35% / 35% / 15% / 15% | Investigator, Project Quality, Benefit, Feasibility |
Eligibility essentials for DECRA applicants
Start by verifying key dates and interruptions to ensure your fellowship candidacy is valid.
Core rule: to be eligible at the closing date your award of PhD must be on or after 1 March 2019, or be equivalent once allowable career interruptions are applied.
Do this early. Clarify any interruptions—parental leave, illness, or part‑time work—and gather supporting evidence well before submission.
Practical checks and good governance
Document each interruption and its dates so institutional reviewers can confirm equivalence. Register the candidate in RMS ahead of internal cut‑offs to avoid administrative delays.
Use part‑time options strategically: three years full‑time salary can extend to six years if you propose part‑time arrangements to manage caring or other commitments.
- Frame your early career record to show momentum and growing independence while acknowledging mentorship.
- Explain how you will conduct research safely and ethically under your host’s governance and training.
- Include clear learning and leadership activities that develop supervision and management capability over the award years.
Keep a clean, auditable trail of dates and documents. Align eligibility narratives with internal compliance checks so you solve issues ahead of submission rather than under time pressure.
Key dates and the past timeline to learn from
A tight timeline is the backbone of any successful grant—map dates backward from the anticipated announcement. Use the DE25 dates as fixed anchors and build a two‑year outlook so you can protect critical windows across the award years.
From guidelines release to anticipated announcements
Important dates: Guidelines and RMS opened 12 October 2023. Final drafts due to the Research Grants Team by 16 November 2023. ARC close was 7 December 2023. Rejoinders ran 28 March to 15 April 2024. Anticipated announcements occurred 2–13 September 2024.
Rejoinders give applicants a formal chance to answer assessor queries before final decisions.
Internal milestones, RMS submissions, and rejoinders
Treat the RMS upload as a project task with version control, sign‑offs and clear signatories. Lock drafts of the project description, budgets and institutional statements before the Research Grants Team deadline.
- Use the Request Not to Assess window to manage conflicts within program rules.
- Factor review streams—internal peer review, faculty checks and compliance—into your schedule.
- Keep a live register of research dependencies: data, ethics and partner letters aligned to each date.
- Build rapid rejoinder workflows and learning loops so responses are evidence‑based and timely.
Practical tip: rehearse full submission readiness two weeks before institutional lodgement to close gaps early and keep your team aligned.
decra computer science template
Open by summarising how your knowledge base maps to specific research areas and the practical problem your project will solve. Keep this to one page so assessors see fit between investigator experience and project scope at a glance.
Write aims as testable hypotheses and list methods that are efficient, scalable and realistic within available funding and machines. For each aim, define inputs (data), outputs, risks and clear success measures.
“Assessors reward clarity: link workpackages to milestones, learning goals and measurable outputs.”
Provide systems diagrams showing data flows, model components and evaluation loops. Map workpackages to a three‑year timeline and show how the phd candidate gains leadership and independence.
- Specify hardware and software with cost justification tied to feasibility.
- Align budget lines to research‑intensive phases and risk mitigation.
- Use ARC headings—Investigator, Project Quality, Benefit, Feasibility—to mirror review forms.
| Assessment area | Weighting | Funding | Example deliverable |
|---|---|---|---|
| Investigator / Capability | 35% | Salary $112,897 pa | PhD supervision plan, leadership milestones |
| Project Quality & Innovation | 35% | Up to $50,000 pa project funds | Systems prototype, reproducible experiments |
| Benefit | 15% | Project travel & engagement | Policy brief, industry pilot |
| Feasibility | 15% | HPC / data access | Risk register, contingency plan |
Translating AI & Cybersecurity ideas into DECRA-ready research aims
Frame aims as focused, testable hypotheses that a three‑year phd can complete. Design each aim around a narrow anomaly detection problem in a real‑world setting so outcomes are measurable and persuasive.
Choosing research areas
Prioritise machine learning methods that suit signal levels and label quality. Use deep learning when complex feature learning improves detection beyond simpler models.
Linking to industry applications
Match aims to clear applications: critical infrastructure monitoring, health data protection, or small‑business security pilots. Show when and how industry partners will provide datasets, evaluation access or deployment trials.
“Deliver demonstrators at 6–9 month intervals to keep momentum and prove translation.”
- Data and systems: describe ingestion, preprocessing, benchmarks and ablation tests that validate robustness, fairness and reproducibility.
- Development sprints: schedule demonstrators with metrics for each sprint and a plan to iterate on failure modes.
- Candidate leadership: assign the phd candidate primary ownership of experiments, with a supervision plan that builds independence.
- Ethics and materials: use public datasets and red‑team scripts with approvals, documentation and safe handling.
- Justify choices: explain why a given model‑system balances compute efficiency, reproducibility and long‑term maintainability.
Align each aim to the program criteria by tying methods and milestones to investigator capability, project quality, benefit and feasibility. This shows assessors you can undertake research that is ambitious, deliverable and clearly valuable to industry and the nation.
Scoring high on Investigator/Capability
Demonstrate leadership by framing a clear, evidence‑based story of your research achievements and supervision plans. Start with concise examples that show originality, influence and independent contributions.
Build a narrative that assessors can follow. State how your outputs have led to uptake, citations or practical use. Show how you created training or opportunities for others.
Track record, leadership and supervision potential
Frame your record around three strengths: scholarly outputs, team leadership and stewardship of systems or platforms. Give short examples of mentoring, community code, or workshop organisation that grew the field.
- Translate papers into measurable impact—adoption, policy reference or open‑source use.
- Map supervision plans: co‑supervision roles, phased milestones for a phd candidate, and inclusive lab practices across the grant year.
- Show project stewardship: budgets met, risks managed, and reliable systems maintained for reuse.
“Assessors reward clear links between leadership, outputs and student development.”
Conclude with a forward plan that scales mentorship responsibly and creates lasting opportunities for early career researchers. Align this to the program language and highlight how a supported phd will deliver high quality research and broader benefits for applicants and collaborators.
Project Quality and Innovation in computer science proposals
Begin with a clear statement of how your methods advance knowledge while remaining deliverable in three years. Frame innovation as testable advances that link prior outcomes to bolder next steps without overreaching.
Create new knowledge by defining reproducible artefacts: benchmarks, libraries and pipelines that your phd candidates and peers can reuse. Show preliminary code, pilot datasets and survey results as evidence of readiness.
Create efficient, scalable systems
Design architectures with complexity bounds and resource envelopes that match budget and HPC access. Justify machine and model choices against simple baselines and alternatives.
Position your knowledge base and methods
Specify interfaces, selection rationales and ablation plans that validate novelty beyond incremental tweaks. Embed evaluations that probe generalisation, bias and failure modes, not just headline metrics.
“Tie every claim to verifiable evidence so reviewers see substance behind ambition.”
- Link prior results to next steps that balance ambition and feasibility.
- Plan milestones that deliver citable outputs each year: datasets, reproducible code and benchmarks.
- Use synthetic or privacy‑preserving materials to de‑risk access constraints.
| Aspect | What to include | Why it matters | Example deliverable |
|---|---|---|---|
| Quality | Prior results, clear hypothesis | Shows credibility and focus | Reproducible experiment log |
| Efficient scalable | Complexity analysis, resource plan | Matches scope to budget | Lightweight model with benchmarks |
| Knowledge base | Pilot data, preliminary code | Reduces execution risk | Open dataset and starter repo |
| Learning evaluations | Bias, robustness tests | Demonstrates real-world value | Evaluation suite and report |
Roadmap: connect innovation threads with year‑by‑year milestones that produce field‑shaping outputs while training phd researchers in reproducible development and leadership.
Designing Benefit for Australia and clear impact pathways
Plan how each research output will deliver tangible value for Australian communities and industry. Start with short, verifiable pathways from prototype to adoption and state who benefits and when.
Economic, environmental, social and cultural value
Translate technical wins into measurable national gains. Describe reduced cyber loss, stronger small businesses, safer critical systems and a growing skilled workforce.
- Align language to the Australian Government priorities and value for money.
- Define real world adopters: industry partners, government agencies and community groups.
- Map outputs to short‑ and long‑term milestones: documentation, pilots, workshops and sector rollouts.
Show how the project trains phd trainees and builds transferable skills for the digital economy. Budget explicitly for adoption activities so impact is driven, not assumed.
“Use case studies and clear metrics — uptake, policy references and vulnerability reductions — to make the benefits credible.”
Identify beneficiary streams and track progress with transparent metrics. This gives assessors a clear line from research to lasting national benefits and helps early career researchers scale outcomes confidently.
Feasibility and budget planning that passes scrutiny
Feasibility depends on a tight three‑year plan and transparent budgets. Break the work into sequenced sprints: year 1 for data collection and baseline experiments, year 2 for model development and systems integration, year 3 for evaluation, pilots and dissemination.
Building a realistic three‑year workplan
Include generous buffers and decision gates. Schedule demonstrators every 6–9 months and set go/no‑go criteria that let you stop, pivot or scale based on evidence.
Budget lines: personnel, data, HPC, and research tools
Salary is costed at $112,897 pa (30% on‑costs) and up to $50,000 pa for project funds. Tie each line to tasks: personnel for supervision and experiments, data licensing and storage, high performance computing, cloud credits, and materials and tooling.
Risk management and mitigation
Maintain a risk register with triggers and mitigations: alternative datasets, lighter machine models, and fallback evaluation streams. Use institutional in‑kind support (Intersect HPC, eResearch estimates) to reduce cash costs and show feasible compute paths.
“Align budgets to outcomes so assessors see clear cost-to-impact logic.”
- Data governance: licensing, storage tiers, backups and sharing policies.
- MLOps: reproducible pipelines, CI for experiments and deployment plans.
- Roles: candidate owns experiments; collaborators supply datasets and domain checks.
| Item | Allocation | Purpose | Outcome |
|---|---|---|---|
| Salary | $112,897 pa | PhD candidate & leadership time | Deliverable milestones, supervision |
| Project funds | $50,000 pa | Data, materials, cloud, tools | Prototypes, pilots, reports |
| HPC / IT | Mix of in‑kind & paid | Model training & benchmarks | Reproducible experiments |
| Risk reserve | Contingency 8–10% | Alternative datasets & evaluation | Mitigated delays |
Summary: tie every cost to a task, use decision gates to protect time and funds, and document governance so you can confidently conduct research and deliver national benefit.
Collaboration that counts: national and international partners
Strong partnerships multiply impact. Build relationships that bring new methods, realistic data and credibility beyond what’s available locally. Aim for collaborations that clearly advance the project and support phd training.
Leveraging industry and government opportunities
Engage industry early to define problems, share traces and create deployment paths. Formalise roles so the team can move fast while applicants lead the research agenda.
- Co‑design pilots with industry partners to test solutions in real settings.
- Use MOUs and short letters to show commitment without over‑promising.
- Plan exchange visits and internships to boost phd learning and practical skills.
Aligning with Australian Government priorities
Connect partnerships to programs and standards bodies that scale reach. Show how the project supports national security, productivity and sovereign capability.
“Partnerships that map to government priorities strengthen impact claims and grant credibility.”
| Collaboration type | Role | Benefit |
|---|---|---|
| Industry partner | Provide datasets, pilot sites, internship places | Realistic evaluation, deployment pathways |
| Government agency | Policy guidance, standards alignment | Pathway to national adoption and impact |
| International lab | Methods exchange, comparative benchmarks | Credibility, broader validation |
| Standards & testbeds | Shared platforms, certification routes | Multiplicative reach, reuse |
Design joint milestones: move from scoping to piloting to evaluation. Define governance for IP and publishing so collaborations evolve across streams and keep phd supervision central to learning outcomes.
University support services to strengthen your application
Turn back‑end support into front‑line evidence by naming the services, timelines and costs that make your proposal credible. A short, verifiable support plan reassures assessors that the project is deliverable and well governed.
Library impact services and research data management
The Library’s Impact Service can produce field‑normalised metrics and narrative evidence to back performance claims. Use their reports as independent corroboration of outputs and influence.
Co‑design a research data management plan with library staff to meet ethics, retention and open access requirements. This shows you can handle sensitive material and share outputs responsibly.
eResearch: HPC estimates, storage, platforms, and in‑kind
Engage eResearch early for high performance compute estimates and in‑kind calculations (for example, Intersect contributions). That advice stretches budgets and clarifies capacity needs.
Choose storage and compute platforms that match institutional environments and security rules. Add internal memos or letters that confirm availability and costs for the fellowship period.
“A cohesive infrastructure story strengthens feasibility and helps your phd team plan milestones with confidence.”
- Use program‑aligned templates for DMPs and costing to speed approvals.
- Align workflows to institutional tools for version control and reproducibility.
- Include phd training on data stewardship and secure coding as a funded activity.
| Service | What they provide | Why it matters |
|---|---|---|
| Library Impact Service | Metrics, narratives, OA advice | Evidence for investigator capability |
| eResearch | HPC estimates, storage, platform advice | Realistic costings and in‑kind support |
| Research Office | Program templates, letters, compliance checks | Streamlines approvals and strengthens feasibility |
For additional guidance on fellowship supports, link supporting materials such as the faculty fellowships brochure when referencing institutional arrangements.
From template to submission: RMS workflow and rejoinders
A calm, repeatable RMS rhythm removes avoidable errors and keeps your project narrative aligned across application fields.
RMS user guides and best‑practice submission rhythm
Follow the RMS user guides step‑by‑step: requesting accounts, submitting applications and lodging Requests Not to Assess. These guides reduce account and certification errors and save time on the final day.
Build a submission cadence: schedule internal rehearsals, one full dry run four weeks out and a final rehearsal 72 hours before institutional cut‑off. Use version control for years of track record and link ORCID records to ensure accuracy.
- Keep aims, significance, benefit and feasibility consistent across all RMS sections.
- Use mentor, peer and compliance streams to check clarity, tone and eligibility.
- Attach only necessary materials and follow file naming, format and page limits.
Prepare rejoinder templates in advance. Rejoinders let applicants respond to assessor comments; diarise the rejoinder window and assign roles for drafting, review and sign‑off.
“Rehearse responses and focus on concise, evidence‑based replies rather than defensive rebuttals.”
Finally, create a 72‑hour playbook: backups, a contingency uploader, final checks and an authorisation chain. Capture learning from dry runs so each submission becomes smoother and more confident.
Broader funding and training landscape to build your case
Build wider support for your proposal by mapping linked scholarships, industry fellowships and training centres that bolster delivery.
Align your phd plan with national schemes to show clear training pathways and resource access. Examples include the CSIRO Industry PhD (iPhD), ARC Training Centres and industry‑backed scholarships that offer tuition waivers and living stipends.
PhD scholarships, Industry PhD, CSIRO iPhD, and Training Centres
These programs provide supervision depth, specialist labs and placements that strengthen feasibility.
- Use industry partnerships (for example, ARC Industry Fellowship PhD scholarships) to secure materials and specialist facilities.
- Offer candidate internships and co‑supervision to accelerate translation and employability.
- Position cohorts in machine learning and cybersecurity as talent pipelines for your team.
Positioning discovery project pipelines and funded research streams
Map how your project can seed or join funded research streams. Show milestones that align grant timelines with scholarship rounds and training centre calendars. Highlight in‑kind materials, shared systems and data resources that reduce risk and increase impact.
| Offering | Benefit | Typical support |
|---|---|---|
| CSIRO iPhD | Industry integration, four‑year training | Tuition waiver, stipend, placements |
| ARC Training Centre | Multi‑partner facilities and cohorts | Access to labs, collaborative projects |
| Industry PhD / Fellowships | Real datasets, deployment pathways | Materials, internships, co‑supervision |
Resources and exemplars to model excellence
Anchor every choice in the official guides and public exemplars. Use factsheets, the Discovery Fellowships Grant Guidelines and sample forms to make sure each claim fits program rules and assessor expectations.
Authoritative sources to start with
Gather the ARC DECRA webpage, the DE25 Instructions to Applicants, the DE25 Sample Application Form and RMS User Guides. These materials show what the scheme provides and how applications are judged.
Build a compact library and reusable systems
Create a versioned folder of key materials, exemplar applications and assessor notes. Keep short modules for Benefit, Risk and Impact so you can reuse text without copying.
- Track the Summary of Changes to stay current.
- Study public LaTeX and form examples to model structure, not wording.
- Plan development sprints aligned to guideline clauses for checks and sign‑offs.
“Use official resources as your foundation; exemplars teach logic and evidence patterns, not language.”
These steps help applicants turn knowledge into polished applications and support learning and development across the team — giving your grant the best chance of success.
Conclusion
Finish with a concise promise: who benefits, when, and how your project will deliver real outcomes.
Keep the research story tight: state the problem, list methods, and name clear measures reviewers can track. Show how your phd candidate gains skills, independence and leadership across three years.
Translate machine learning advances into real‑world outcomes, from anomaly detection pilots to resilient systems and applied workflows. Tie equipment, materials and data choices to each milestone so feasibility and quality are evident.
Close strong: commit to measurable benefits for industry, policy or communities, map short delivery streams, and submit with confidence.