This project focused on building and testing a practical AI security assessment lab for evaluating LLM defenses against prompt injection and jailbreak attacks.

I integrated Spikee by Reversec with a locally hosted cybersecurity model running through LM Studio, then added NVIDIA NeMo Guardrails to compare model behavior under three conditions: no guardrails, input filtering, and combined input/output protection.

The work included configuring the local model environment, building a custom FastAPI gateway, integrating NeMo Guardrails, troubleshooting model latency and timeout issues, creating a reusable Spikee target, and analyzing attack results using Spikee’s built-in reporting tools.

The project also explored different adversarial testing approaches, including prompt injection datasets, obfuscation, encoded attacks, Best-of-N testing, synthetic canary leakage tests, and structured benchmark comparisons.

The objective was to measure how much the guardrails reduced successful attacks while keeping the model, dataset, and testing conditions consistent.

Tools used: Spikee, NVIDIA NeMo Guardrails, LM Studio, Python, FastAPI, PowerShell, Parrot OS, local LLMs, JSONL datasets, and custom security testing scripts.

This project demonstrates a hands-on approach to LLM red teaming, AI safety testing, prompt-injection assessment, and guardrail validation for organizations deploying generative AI systems.

pretty cool tbh. testing the same attacks with different guardrail setups makes the results way more interesting.

Visit the live app here: https://paws-and-puddle-pro.lovable.app/ Meet Sam. He runs a highly requested mobile pet grooming van in Denver, but standard booking calendars are quietly killing his margins. Because generic apps let customers pick any time slot they want, Sam spends two to three hours a day zig-zagging across the city in traffic, burning fuel and losing out on actual appointments just to accommodate a messy schedule.

To fix this, I built Paws & Puddle, a web app that solves his geographic routing nightmare before a booking ever hits the calendar. Instead of a basic form, I engineered a zero-click geo-gate. It automatically fetches a user's zip code via location services and dynamically filters the calendar to only show availability for the specific days Sam’s van is already assigned to their neighborhood.

Because forced logins instantly kill conversion rates, I designed a completely frictionless guest checkout. But to ensure long-term retention, the Thank You page features a "Guest-to-Auth" funnel that seamlessly converts their inputted data into an account post-booking, simply by asking the user to "set a password to track your van's ETA."

Backed by a full operator console where Sam can manage custom service zones, assign route days, and trigger automated status emails, this build proves that premium, minimalist design and aggressive problem-solving belong in the exact same app.

Check out the demo video below to see the end-to-end flow in action! @Lovable #lovablechallenge

Very nice visual design. I love the dog drawings 😍