# Maryam Touimi Benjelloun

> Maryam TB is a software engineer, creative technologist, and founder of FURTHER PROJECTS. This is her personal portfolio, covering selected software projects, experience, and experiments.

Maryam developed the original WindowSwap application and previously worked at Twitter, Zero Studios, and Serif (now Canva). She is an Awwwards developer jury member. Current work includes Rook and an NL2SQL AI application for Baker Tilly Seven, connected to a data lakehouse via MCP.

## Pages

- [Home](https://maryamtb.com/): Introduction and current work.
- [About](https://maryamtb.com/about): Biography, experience, technical skills, and education.
- [Work](https://maryamtb.com/work): Selected projects, technology stacks, results, press coverage, and project links.
- [Lab](https://maryamtb.com/lab): Experiments and explorations.

## Selected projects

- [Baker Tilly Seven](https://maryamtb.com/work): NL2SQL application for conversational analytics, connected to a data lakehouse via MCP. Technologies include LangGraph, FastAPI, React, TypeScript, PostgreSQL, and Redis.
- [Rook](https://userook.app/): A native macOS notes app for code, built with Swift, SwiftUI and AppKit. Tree-sitter syntax highlighting, on-device AI with Apple’s Foundation Model or BYOK, and local agents can save content directly to Rook via MCP. [Source code](https://github.com/maryamtb/rook).
- [WindowSwap](https://window-swap.com/): Maryam was the principal developer of the original React application, using the Vimeo Player SDK. The Work page includes linked press coverage and recognition.
- [CorpAI](https://corpai.io/): Built an enterprise AI gateway with 230+ MCP servers and an AI orchestrator integrated into Slack, Google, and web chat. Implemented RBAC, custom policies, and audit logging for secure tool access. Architected separate control and data planes for hosted and BYOC deployments. Technologies include Next.js, FastAPI, PostgreSQL, Redis, Kubernetes, Terraform, and AWS.
- [archo](https://www.getarcho.com/mobile): Collaboration platform for architecture firms, later expanded into an AI homebuyer assistant. Technologies include React Native, Expo, NestJS, PostgreSQL, pgvector, Redis, AWS, LangChain, and MCP.

## Lab projects

Project previews are available on the [Lab](https://maryamtb.com/lab). Decoder’s full project notes are available; other detail pages are not published yet.

- [Decoder](https://maryamtb.com/lab/decoder): Graph-based Python code navigation for engineers and agents.
- Clarity pt. I: ML-powered sound detection and adaptive audio. Wearable concept.
- AI Rover: Voice-to-action robot control using natural-language instructions.
- Swave: EEG headband and app concept for real-time brainwave interpretation.
- Echo: An AI creative therapist based on creative research.
- CBCR: Exploration of cigarette-butt collection with Arduino, ultrasonic sensors, and computer vision for targeted object identification.
- Llama fine-tuning: LoRA fine-tuning to specialize a coding agent in LangChain.

## Contact and profiles

- [Email](mailto:hello@maryamtb.com): Contact Maryam.
- [FURTHER PROJECTS](https://further-projects.com): Maryam's design and technology studio.
- [GitHub](https://github.com/maryamtb): Public repositories.
- [LinkedIn](https://www.linkedin.com/in/maryam-tb): Professional profile.
- [X](https://x.com/mimobenjo): Social profile.