AI Software Engineer (Fullstack)

Clearstory•Walnut Creek, CA

1d•Hybrid

About The Position

Clearstory is hiring an AI Agent Engineer (Fullstack) to help build the next generation of intelligent automation across our platform. AI agents are becoming core to how Clearstory delivers value — surfacing change order insights, automating cost workflows, and giving construction teams superpowers they didn't have before. In this role, you'll own AI agents end-to-end: from scoping a workflow with product, to designing prompts and tools, to building the eval harnesses that prove the agent works, to shipping the surfaces where the agent shows up in our product. This is a high-impact role for an engineer who's been building with LLMs in production and is ready to own a category-defining surface area. You'll work directly with our Head of AI, product leaders, and senior engineers to ship agents that move the needle for general contractors, specialty contractors, and owners managing billions of dollars in change orders every month. Clearstory is an AI-forward engineering organization. We use modern AI tools every day — not as a novelty, but as core leverage. You'll have access to the best agent platforms, frontier models, and internal tooling we've built to ship faster while holding a high bar for craftsmanship and reliability. We operate in a hybrid work model: 3 days per week in the office, with the option to work 2 days per week from home if desired. We value in-person collaboration while supporting flexibility.

Requirements

Has shipped production LLM systems — not prototypes, not hackathon projects. You can talk through an agent or workflow you've built, including its failure modes and the evals you put around it.
Operates with high agency in ambiguous environments. You don't need a PRD to start moving.
Is genuinely fullstack — comfortable from Postgres up through React, and not allergic to either end. Bonus points fo Golang experience.
Communicates clearly across engineering, product, and design. Low ego, collaborative, direct.
Cares deeply about reliability and craft. AI quality, latency, and cost tradeoffs are second nature.
Thrives in a fast-paced startup environment where the surface area is wide and the leverage is real.
Embodies our core values: Be Curious, Customer Obsession, and Keep It Simple.
2-3+ years of professional software engineering experience, including production work on AI/LLM systems.
Strong proficiency in Python and TypeScript, including async programming.
Fullstack production experience — React/TypeScript on the frontend, modern backend services, REST APIs, Postgres.
Hands-on experience with modern LLM tooling: prompt engineering, function/tool calling, RAG patterns, vector stores, and at least one major model provider (Anthropic, OpenAI, Google).
Experience designing and running evaluation frameworks for LLM systems.
Comfort with cloud infrastructure (GCP preferred; AWS or Azure acceptable), Docker, and CI/CD.
Strong written and verbal communication — you can scope an agent with product, explain tradeoffs to an exec, and write docs your teammates actually want to read.

Nice To Haves

Experience with agent orchestration frameworks (LangGraph, CrewAI, Claude Agent SDK) and observability tooling (Braintrust, LangSmith, Datadog, or equivalent).
Familiarity with MCP (Model Context Protocol) and building MCP servers or skills.
Experience with BigQuery, ClickHouse, or similar analytics warehouses.
Previous work in ConTech, FinTech, or other regulated enterprise verticals.
Background as a founding engineer or in an applied AI / platform engineering role at a fast-moving startup.

Responsibilities

Own AI agents and workflows end-to-end — from scoping the workflow with product and design, through prompt and tool design, to production deployment and continuous iteration.
Build across the full stack — agent orchestration backends, tool/API integrations, evaluation pipelines, observability, and the product surfaces where agents show up.
Design and run evals — golden sets, LLM-as-judge frameworks, regression evals — so we ship agents that work, not demos that don't.
Integrate with the construction tech ecosystem — accounting systems, project management tools, document repositories — and our own BigQuery-backed data platform.
Partner closely with product, design, and other engineers to translate messy real-world workflows into reliable agent behaviors.
Shape the future of our agent platform — surface unmet needs, prototype new patterns, and influence how AI shows up across Clearstory's product.
Move fast in a SOC 2–compliant environment, using modern AI development workflows.