Software Engineer, AI Systems & Infrastructure - AI Geospatial Assistant Team

Descrição da Vaga

Requisitos

• Bachelor’s Degree in computer science or an equivalent field
• 4+ years of experience building and scaling high-performance backend systems in Python, Go
• Proficiency in AWS or GCP, including experience with Infrastructure as Code (Terraform) and CI/CD pipelines for high-availability services
• Solid familiarity with LLM orchestration frameworks (LangChain, LlamaIndex) from an implementation perspective—knowing how to build reliable agents, handle retries, and manage state
• Proficiency with vector search pipelines and high-performance distributed computing
• Experience with distributed computing frameworks like Docker, Kubernetes, or Ray Serve
• A track record of applying scientific rigor to "ground truth" verification in AI models to maintain user trust and credibility

Responsabilidades

• As a Software Engineer, you will help to build the backend systems that bring our AI Geospatial Assistant to life
• While our research teams develop the core models, you will be responsible for the 'last mile' of delivery, architecting the high-throughput backend services, scaling our systems, and ensuring our agentic workflows are fast, reliable, and cost-effective at a global scale
• This is a full-time, hybrid role which will require you to work from our San Francisco office 3 days per week
• Develop and optimize multimodal LLM applications
• Work with and support the infrastructure needed for scaling and delivering embeddings
• Architect AI Orchestration: Build and maintain the high-scale systems required for LLM orchestration and agentic workflows, ensuring low-latency responses across petabytes of imagery
• Operationalize Research: Collaborate with backend engineers to transition experimental AI models into stable, low-latency production inference endpoints
• Build AI Observability: Implement production-grade monitoring, logging, and tracing for our AI services to ensure reliability and facilitate rapid debugging of our systems
• Benchmark Performance: Define model success criteria and instrumentation to ensure the assistant consistently outperforms vanilla LLM alternatives

Benefícios