AI Solutions Developer
Building AI-enabled web systems, agent infrastructure, and developer automation. AI Solutions Developer at WEBINFORM IT Ltd and BSc Artificial Intelligence student at Vrije Universiteit Amsterdam.
- 🎓 BSc Artificial Intelligence student at Vrije Universiteit Amsterdam
- 🔧 AI Solutions Developer working on AI integrations, backend logic, automation, and user-facing AI features
- 🧭 Interested in roles that combine technical implementation, product thinking, and project delivery
- 📍 Based in the Rotterdam area, Netherlands — available for full-time work
- 🛠️ Working primarily with Python and TypeScript, with public projects in Rust
| Project | Implemented scope | Stack |
|---|---|---|
| VoidArch Context | Principal flagship: local-first repository memory, search, code graph, durable project state, and token-budgeted context packs for AI coding agents | TypeScript · SurrealKV · Tree-sitter · ONNX |
| VoidArch Studio | Active-development control room for coding-agent sessions with daemon-owned PTYs, worktree and session APIs, a localhost dashboard, deterministic safety hooks, and a Tauri desktop shell | TypeScript · Node.js · Rust · Tauri · xterm.js |
| AgentSec Suite | Deployed AI-agent security governance suite combining deterministic runtime inspection, approval queues, PromptShield, MCP Guard, AgentMap, integration guides, and an optional copilot interface | Next.js · TypeScript · Prisma · Vitest |
| saas-core | Supporting modular SaaS factory with a module registry, presets, environment planning, generated contracts, dry-run scaffolding, and a reference application | Next.js · TypeScript · Prisma · GitHub Actions |
Interactive, deterministic demos on davidpapp.dev (no LLM, no accounts). Personal projects stay prototypes — not production systems.
| Demo | What it is |
|---|---|
| PromptShield / AgentSec | Live rule-based prompt-injection scanner (the public AgentSec surface) |
| Self-Interview | The site interviews David from verified facts only; you can jump in |
| Task-to-Flow | Describe a repetitive task → an illustrative automation flow. Hours are assumption sliders, not measured results |
| Fit-Finder | 3-tap quiz (industry × intent × urgency) → an illustrative role lens |
Self-Interview, Task-to-Flow, and Fit-Finder ship on /roles/<id> and /demos/<slug>. Company links from applications use /?c=<company> (audience framing only — the page does not invent facts about the company).
Recent private lab work — source on request, not claimed as production:
- Unclickbaiter — transcript-grounded YouTube title rewrite + thumbnail-honesty spec (heuristic or local LLM; no paid APIs)
- CS2 sticker craft optimizer — search sticker placement/scale/rotation/wear; heuristic or local VLM scoring
- Active Companion — local meeting listener → retrieval cheat-sheet UI (citations or explicit gaps)
Languages
AI & Agent Systems
Web, Backend & Data
