Role OverviewMatter is hiring a Telemetry and Observability Engineer to make datasets, training runs, models, environments, agents, and missions observable as one AI system. Reporting to Ignacio Cases Martin, this individual contributor will connect flight and mission reality with infrastructure, model, and product behavior through shared telemetry, traces, alerts, reliability mechanisms, and operational context.
What You Will Do
Build a common telemetry model linking mission state, datasets, code, checkpoints, accelerators, model and prompt versions, environment episodes, agent steps, tools, decisions, cost, latency, and feedback. Instrument distributed training, data loading, GPU utilization, checkpointing, experiment health, batch and online inference, and model-serving behavior.
Why It Might Be a Fit
Teams can correlate mission, data, infrastructure, model, environment, and agent behavior during normal operation and incidents. Alerts and service objectives identify actionable failures without obscuring scientific or operational context. Capacity, cost, version, and reliability signals support safe deployments and faster diagnosis across the AI stack.
Requirements
- Experience with ML observability, telemetry, time-series systems, distributed training or inference, cloud infrastructure, or production AI platforms.
- Strong hands-on engineering skills in observability, automation, APIs, infrastructure, incident tooling, and systems debugging.
- Understanding of metrics, logs, traces, events, model and data monitoring, service objectives, capacity, release safety, and incident diagnosis.
- Ability to reason about AI failure modes including bad datasets, training instability, stale checkpoints, drift, version mismatch, environment bugs, and agent-tool failure.
- Experience designing telemetry schemas and correlation workflows across multiple systems, time domains, and ownership boundaries.
Benefits
- Competitive compensation based on experience
- Early-stage equity package
- 100% employer-paid health, dental, and vision coverage
]]>