Responsibilities

Hardware programs generate far more telemetry than any team can review by hand. Sift Agents is our answer: AI agents that work alongside hardware engineers, investigating anomalies, reviewing test and flight data, and answering in minutes questions that used to take a day of scripting.
This role comes with real product ownership and autonomy. You will talk directly to customers to understand how they review their machines, decide what to build, and own it from first MVP through every iteration after it ships. The surface area is wide: one day you might be polishing the frontend micro-interactions that make the chat experience feel great, the next you might be writing the Kubernetes scaling logic that orchestrates agent workloads. We’re heavy users of frontier agentic products ourselves and have strong opinions about what makes them great. Sift Agents is where we put those opinions to work.
This is a product engineering role, not a research role. Most of the work is the hard part of making agents dependable: tool design, context management, code execution environments, evaluation, and observability.
Talk directly to customers and partner with product to turn real review workflows into agent capabilities: generating dashboards, writing analysis scripts, and surfacing insights buried in their telemetry
Design, ship, and operate agentic systems that reason over large-scale time-series data and hardware domain context
Build everything around the model that makes agents dependable: tool interfaces, sandboxed execution for agent-generated code, context and memory management, custom compaction algorithms, opinionated skills, and guardrails
Develop and maintain Sift’s MCP server, the tool surface that lets both our agents and our customers’ AI tools query telemetry directly
Work across the whole stack
Build the frontend surfaces where Sift Agents live, and make them feel great
Design and implement the APIs that power our agentic capabilities
Run agents reliably at scale, both in the cloud and on-prem
Build evaluation suites that measure whether agents actually help engineers, and instrument quality, latency, cost, and failure modes in production

Software Engineer, Full Stack (Agentic AI) at Siftstack · LumiJobs

Software Engineer, Full Stack (Agentic AI)

Responsibilities

Qualification

Required

Preferred

Benefits

Company

Siftstack

More roles at Siftstack