Germain Safari
Software Engineer and Machine Learning Nerd
New to Contra
Germain is ready for their next project!
Engineered and maintained Azure cloud environments supporting DevOps, identity, and machine learning workloads. Built and managed CI/CD pipelines, automated deployments, configured infrastructure and access policies through Microsoft Entra ID, and supported ML lifecycle workflows using Azure ML and Azure AI Foundry. Helped deliver secure, scalable, and production-ready cloud solutions for application and AI services.
0
5
OpenClaw as an agentic AI system for autonomous task execution, tool use, workflow orchestration, scheduling, and multi-step reasoning. The project focuses on building a reliable AI agent that can manage prompts, skills, memory, automation, cron-based tasks, and integrations across products and operations. It includes design, testing, iteration, and deployment of scalable agent workflows for real-world business use.
0
12
AI voice calling system using Vapi and Twilio. The project connects a compliant Twilio phone number or SIP route to a Vapi assistant, enabling inbound and outbound voice calls with real-time conversation handling. It includes number provisioning, regulatory bundle setup for Poland, SIP or Twilio integration, call routing, testing, and deployment for production-ready voice automation.
0
19
AI agent workflows with Claude as a digital coworker for research, writing, analysis, skill creation, task automation, cron jobs, scheduling, tool use, memory, orchestration, and multi-step execution. The project focuses on creating reliable agentic systems that can plan, reason, delegate, trigger scheduled actions, and support day-to-day business operations through reusable, scalable AI workflows.
0
18
By combining reference images of Usain Bolt, a Formula E car, and Miami architecture, the AI handles 90% of the heavy lifting. Learn how to use multi-image composition to maintain detail and logos, iterate for the perfect variation, and add final polishes in Photoshop.
0
17
Built an image-to-image translation system using Pix2Pix, training a conditional GAN on paired image datasets to generate transformed outputs from input images. Managed the full workflow from data preprocessing and model training to evaluation and deployment. Integrated the trained model into a web application, enabling users to upload images and receive real-time or near-real-time generated results through an interactive interface. The project demonstrated practical experience in deep learning, computer vision, and deploying ML models in user-facing applications.
0
23