We're putting together a dedicated delivery pod to build and run an internal agent platform on AWS Bedrock AgentCore for a global life sciences client. The pod works as one team with the client's engineers to deliver the platform other teams will build their agents on. In this role you make the platform visible and dependable. You'll build the OpenTelemetry instrumentation, observability integrations, tracing, SLOs, alerting, and cost monitoring for token and compute spend. When an agent is slow, wrong or expensive, your work is how the team finds out and fixes it. This is a named-team engagement, so the person we propose is the person who starts. ROLE RESPONSIBILITIES • Instrumentation: Build OpenTelemetry instrumentation standards for agents, tools and platform services, and make it easy for teams to adopt. • Observability Integration: Connect AgentCore Observability and CloudWatch with the client's existing observability tools (e.g., Datadog, Splunk, Grafana). • Tracing: Set up end-to-end tracing across agent steps, model calls, tool calls and multi-agent handoffs so issues can be traced to their source. • SLOs and Alerting: Define SLOs for latency, availability and error rates, and build alerting that is useful and not noisy. • Cost Monitoring: Track token usage and compute spend by team, agent and environment, with dashboards, budgets and alerts for unexpected spikes. • Incident Readiness: Write runbooks, support incident response and run post-incident reviews that lead to real fixes. QUALIFICATIONS • 6+ years in SRE, platform reliability or observability engineering, with strong hands-on AWS experience. • Hands-on experience with OpenTelemetry (SDKs, collectors, exporters) and distributed tracing. • Experience with Amazon CloudWatch and at least one enterprise observability platform (Datadog, Splunk, Grafana, New Relic or similar). • Experience defining and running SLOs, error budgets and alerting strategies. • Experience building cost visibility and FinOps reporting on AWS (Cost Explorer, CUR, tagging strategies). • Scripting and coding skills in Python, Go or TypeScript. • Clear communication and able to turn data into decisions for technical and non-technical audiences. PREFERRED QUALIFICATIONS • Experience monitoring LLM or agent-based applications, including token usage, latency and quality signals. • Familiarity with LLM observability tools (e.g., Langfuse, Arize, LangSmith) or AgentCore Observability. • Experience with infrastructure as code (Terraform or AWS CDK) for monitoring resources. • Experience in pharma, life sciences or another regulated industry. WHO YOU ARE • You want to know what's happening in a system before a user tells you • You build alerts people trust and dashboards people actually open • You treat cost as a reliability concern, not an afterthought • You stay calm in incidents and focus on learning afterward • You work well inside a client team and build trust quickly • You have prior experience in consulting • Prior experience and connections in the Life Sciences industry is preferred