Fredericton, NB • youneslaaroussi.ca • github.com/youneslaaroussi

I am a Computer Science and Mathematics student at the University of New Brunswick (4.15 GPA) specializing in high-performance systems and AI infrastructure. I focus on building the low-level "guts" of software—from Win32-native C++ recording engines to real-time threat detection pipelines at the Canadian Institute for Cybersecurity.

I approach engineering with a founder’s mindset: optimizing for scale, performance, and measurable technical leverage.

• Current: Software Engineer @ Canadian Institute for Cybersecurity (CIC)
• Status: Government of Canada Reliability Status Clearance

High-performance desktop video stack & OS-level recording engine.

• Low-Level Core: Engineered a Win32-native recorder in C++ using N-API to bridge DXGI desktop duplication and NVENC/NV12 hardware encoders.
• Performance: Implemented zero-copy ring buffers to share D3D11 textures with the Electron renderer, maintaining sub-frame latency.
• Architecture: Orchestrated a multi-process Chromium/FFmpeg export pipeline for high-fidelity ProRes/H.264 output.

CLI-driven carbon intelligence for GitLab CI/CD.

• Infrastructure: Developed a CLI that instruments GitLab runners to calculate CPU/RAM emissions using live CloudWatch and Electricity Maps data.
• Automation: Built automated .gitlab-ci.yml patching with encrypted credential storage and PostgreSQL emissions baselines.

🏆 1st Place Overall — Google + Square AI Hackathon

• Engineering: Augmented Square databases with AI-driven psychographic analysis and voice-enabled chart explanations.
• Stack: Google Vertex AI, Square APIs, Next.js.

🏆 Winner — 1Password Hackathon

• Contribution: Built the official PHP/Laravel bindings for 1Password’s Passage SDK, enabling seamless biometric authentication for PHP environments.

Software Engineer | Jan 2025 – Present

• Inference Optimization: Slashed threat detection latency by 65% by refactoring FastAPI endpoints and optimizing Elasticsearch queries.
• AI Security: Deployed Qwen LLMs for automated reporting with custom guardrails to mitigate prompt injection attacks.
• Data Pipelines: Architected Kafka streams using FastText for real-time language detection with 95%+ accuracy.
• DevOps: Reduced CI/CD runtime by 75% through parallelized GitHub Actions and Docker layer caching.

• 1st Prize: ForestShield: AWS Deforestation Detection (AWS Lambda, SageMaker, Sentinel-2)
• Honorable Mention: Marionette: On-Device Multimodal AI (Gemini Nano, Transformers.js)
• 3rd Place: AutoIR: Agentic Incident Response (TiDB Vector Search, AWS)

• Languages: C++, Python, JavaScript (TypeScript), Java, PHP
• Systems/Cloud: Win32 API, AWS (Lambda, SageMaker), GCP, Docker, Kubernetes
• AI/Data: LLMs (Gemini, Qwen), Kafka, Elasticsearch, PostgreSQL, Vector Search
• Frameworks: React.js, Node.js, FastAPI, Electron, Laravel

• Email: hello@youneslaaroussi.ca
• LinkedIn: linkedin.com/in/younes-laaroussi