AI Engineering Student · VU Amsterdam

David Papp —
AI Engineering Student

2nd-year BSc Artificial Intelligence at VU Amsterdam. I build production-quality LLM tools: agent observability, fine-tuning pipelines, and retrieval systems.

Recently consulted for an AI-first startup to cut LLM API costs by ~40% and harden safety controls against prompt injection. Looking for junior AI/automation engineer or data science roles in the Netherlands.

Email me Book a 20-min call Download CV

Three Pillars

Each project tackles a real problem in the AI engineering stack.

MCP Sentinel

Drop-in observability for agent tool calls. Log, guard, and audit every MCP interaction without changing your tools.

Explore Sentinel →

Custom Training

Codebase to dataset to LoRA in your infra. Turn internal knowledge into fine-tuned models automatically.

Explore Training →

RAG + 3D Chat

Chat with your docs, render 3D outputs. Retrieval-augmented generation with interactive visual context.

Explore Chat →

Quickstart in 3 lines

Reference implementation — this is the API surface I designed for MCP Sentinel. See the architecture section for how it works under the hood.

mcp-server.ts

import { MCPSentinel } from'./sentinel';

const sentinel = new MCPSentinel({ apiKey, guards: ['injection', 'pii', 'cost'] });

// Reference impl — github.com/pappdavid/mcp-sentinel
const server = sentinel.wrap(yourMCPServer);
server.listen(3001);

Build Philosophy

Engineering principles that ship reliable AI products.

Cost-Aware

Every API call has a price tag. Rate limiting, caching, and tiered routing keep costs predictable.

Security-First

Input validation, HMAC signing, RLS policies, and guard rails are built in — not bolted on.

Minimal Surface

Ship the smallest thing that works. Three lines of code beats a premature framework.

Portfolio Assistant

Ask me anything about my work

Ask about my projects, experience, or stack — answers come from my portfolio content.

David Papp —AI Engineering Student