Freelancers using Dust in Zagreb
Freelancers using Dust in Zagreb
Sign Up
Post a job
Sign Up
Log In
Filters
2
Projects
People
Sean Honan
Zagreb, Croatia
AI MVP builder helping founders turn messy ideas into usable
New to Contra
Follow
Message
AI MVP builder helping founders turn messy ideas into usable
1
Scientific Validator: AI Document Integrity Review Tool Scientific Validator is a prototype for reviewing whether a report, paper, or AI-generated document is structurally trustworthy before a human relies on it. The system checks documents across multiple integrity layers: Reasoning structure Source and citation traceability Method consistency Core inquiry continuity Unsupported or overconfident claims Possible fabricated or weakly grounded references Polished language that hides weak logic The product challenge was to turn a complex validation process into a clear, usable interface. Instead of giving one vague “AI score,” the tool breaks the review into separate report cards so the user can see which parts of the document are sound, which parts need review, and where the reasoning may fail. My role covered product architecture, validation logic, AI workflow design, UX structure, report framework, and demo narrative. Key focus areas: AI document review Source-grounded validation Human-in-the-loop review Research/report integrity Structured AI outputs Multi-agent review workflows Clear UX for complex reasoning checks
1
8
1
SEO Blog Builder for Language Schools SEO Blog Builder for Language Schools is a practical internal AI content tool for small education businesses. It helps users generate structured blog drafts for course promotion, lead generation, brand awareness, student retention, community building, and SEO visibility. The MVP focuses on clear inputs, useful output structure, and education-specific marketing logic rather than generic AI writing.
1
5
1
LucidLock: Pre-Execution Integrity Layer for AI Workflow Safety LucidLock is a prototype interface for controlling high-risk automated actions before they execute. The demo shows how an AI-assisted workflow can verify whether an action is still authorised at the final moment: checking the approved instruction, executing agent, policy context, semantic match, replay status, expiry, and live system state before allowing execution. The product challenge was to turn complex backend safety logic into a clear decision interface. The user needs to see what was requested, what changed, which checks passed, which risks were detected, and why the final action was approved or blocked. My role covered product architecture, workflow logic, UX structure, prototype direction, and demo narrative. Key focus areas: AI workflow safety Pre-execution validation Agent identity and authority checks Human-readable decision logic Risk/status dashboard design High-stakes action approval flows Clear interface for complex system state
1
17
1
Child Drawing Reflection Tool - Mobile MVP Flow A mobile-first browser MVP that helps parents and carers respond thoughtfully to young children’s drawings. The tool guides adults through photo upload, visible-feature review, child context settings, and gentle conversation starter generation. The product is deliberately bounded: it does not diagnose, interpret, or analyse the child’s drawing. It focuses on privacy, child-led meaning, clear steps, and calm UX.
1
27
Dust
(1)
Follow
Message
Explore people
