Loqmen Anani

Architected and built from scratch an end-to-end multi-agent LLM system (scriptwriter, illustrator and layout agents) orchestrating 5 AI models and multiple external APIs across 800+ structured steps to generate complete comics with no human intervention. Cut production time from 6 months to under 1 hour and reduced cost 10× , enabling 2,000+ pages sold; owned product strategy and the full production stack.

Developed and trained a deep-learning segmentation model for automated extraction of snake skin scales from high-resolution images. Built a computational pipeline converting 3D scans into standardized 2D representations for quantitative analysis.

Designed and operated a centralized ETL infrastructure on virtual machines, leveraging Google Cloud Platform (VM management, isolation, scalability) for multi-client data pipelines. Developed an internal web application to automate ETL deployment and updates, removing manual SSH operations. Built the backend of a public HubSpot application (OAuth2, APIs, MongoDB) enabling data synchronization with PrestaShop. Automated data workflows and monitoring (mapping generation, pipeline failure alerts), improving reliability and operational efficiency.

Developed a custom Python pipeline to process and analyze high-dimensional multiplexed imaging (CODEX) data with single-cell resolution. Designed spatial analysis algorithms to characterize the neighborhood and interactions of regulatory T cells in human tonsils.

Developed a Python bioinformatics pipeline to analyze ssDNA Capture Hi-C sequencing data, enabling separation of ssDNA and dsDNA contacts.