Senior Software Developer; Applied AI

Position: Senior Software Developer (Applied AI)
Engineering at Clariti runs on our Agentic SDLC framework: agent specs, reusable prompts, orchestration templates, and eval harnesses that let small pods deliver at multiples of traditional velocity
The Applied AI pod owns that framework and is now extending it beyond engineering: into our Professional Services delivery practice and into the Clariti AI harness, the platform of connectors, skills, and guardrails that lets non-technical employees across Sales, CX, Finance, People, and PS use AI safely on real work
You are the senior engineer on this pod
Think of the job as Dev Ops for AI: you do not ship product features, you ship the infrastructure that makes everyone else faster
Where a Dev Ops engineer builds pipelines, golden paths, and observability for code, you build them for AI: the connector layer, the eval harnesses, the orchestration runtime, the cost and quality telemetry, and the guardrails that make agent output trustworthy enough for government software
You will work alongside a Technical Product Manager who owns the roadmap
…
You own how it gets built, and much of what gets built, because on a pod this small the line between architecture and implementation is yours to draw
Build and evolve the Agentic SDLC framework. Design and implement the agent workflows, orchestration templates, and reusable components the build pods run on. Harden what exists, extend what is missing, and keep the framework fast as model capabilities and our delivery patterns change
Build the connector layer. Design, implement, and operate MCP servers and integrations into our core systems so agents and non-technical employees can act on real company data with correct permissions. Treat connectors as production software: versioned, tested, monitored, least-privilege by default
Make evals the backbone. Build and maintain the eval harnesses that score agent output automatically: regression suites for prompts and workflows, quality gates in CI, and the scoring infrastructure that tells us whether a change to a model, prompt, or workflow made things better or worse. If we cannot measure it, we cannot scale it
Own AI observability and cost telemetry. Instrument token spend, latency, eval pass rates, and usage across every production agent workflow. Build the dashboards and alerts that turn “AI is expensive and mysterious” into a managed system with unit economics per workflow
Engineer the guardrails. Implement the permissioning, audit trails, versioning, and output controls that let agent-assisted work stand up in a government context, where an artifact can end up in front of a planning commission. Make the safe path the default path in code, not in policy documents
Ship enablement infrastructure. Build the skill and template libraries, onboarding flows, and self-serve tooling that take a non-technical employee from zero to producing real work with AI, and the feedback loops that route their usage data back into the platform roadmap
How success is measured:
90 days: you know the Agentic SDLC framework end to end, how the build pods use it day to day, where it is strong, and where it breaks, and you are shipping to it without regression to pod velocity. You have taken over operational ownership of the existing connectors and eval harnesses from the founding team, and your first improvement to the framework, scoped with the Pod Lead against what the pods actually need, is live
12 months: the framework is measurably faster and more reliable than the day you joined, with automated eval gates on every production agent workflow and cost and quality telemetry per workflow driving routing and optimization decisions. The connector layer covers our core systems and runs like production infrastructure. The Clariti AI harness…