AI Systems Engineer - Agents & Inference

Practical experience building with AI agents, model APIs and tools

Understanding of how agents interact with applications, manage state and recover from failures

Ability to use coding agents effectively to scope, implement and verify work while retaining ownership of correctness and technical decisions

Experience with tool calling, structured outputs, streaming or MCP

Understanding of how AI workloads run across applications, hosts and accelerated hardware, including remotely hosted environments

Comfort working with Linux, processes, resource allocation and diagnosing failures across the software and compute stack

Application of sound security principles to permissions, credentials and workload isolation

Ability to build reliable execution with clear progress, cancellation and recovery

Hands-on experience deploying computer vision models and working with LLM services

Ability to use representative applications and inference workloads to verify platform correctness on real hardware

Skills to assess accuracy, latency, throughput and resource use

Experience investigating failures across model code, inference SDKs and execution environments

We value demonstrated ability and judgment over particular degrees or certifications

Software Engineering: Python, Rust or another systems language. APIs, databases and native integrations. TypeScript/React and Tauri experience

Workload Orchestration: Job submission, scheduling, quotas, concurrency. Result retrieval and reproducible environments. SDK and driver compatibility management

Security: Authentication, authorization, least privilege. Access revocation, sandboxing, network security. Protection of users' files and data

Cloud Operations (Google Cloud/AWS): Containers, infrastructure as code, CI/CD. Monitoring, incident investigation. Safe deployments, rollback and recovery

Applied ML & Accelerated Computing: PyTorch/ONNX. Image/video processing, detection, classification, segmentation. Calibration, quantization and hardware-aware optimization

Evaluation & Retrieval: Reproducible benchmarks for model quality, agent behavior and task success. Grounded retrieval and traceable results

Fine-tuning, multimodal applications, distributed inference

Desktop packaging

Experience maintaining developer tools