Senior Agentic Software Engineer

WeHawaii is looking for a Senior Agentic Software Engineer to help build our products and to own the AI workflows that run inside them.

The work is genuinely two-sided. Half of it is senior software engineering: shipping features, modeling data, writing APIs, reviewing code, and keeping things stable for paying clients. The other half is agentic engineering: designing agents and automated pipelines that read legislation, summarize hearings, enrich records, draft content, and handle work that used to require a person at a keyboard.

We want someone who treats AI as part of the system architecture rather than a feature bolted on at the end, and who cares as much about reliability on the four hundredth run as about the first impressive demo.

Build and ship product

• Develop and maintain features across WeTrack and WeCampaign, front end through data layer.
• Design schemas, APIs, and background jobs that hold up as data volume grows.
• Keep deployments, monitoring, and error handling boring and predictable.
• Make pragmatic tradeoffs appropriate for a small team serving real clients.

Design and operate AI agent workflows

• Build agents and multi-step pipelines that use tools, call APIs, and act with limited human supervision.
• Choose models thoughtfully, manage prompts and context, and control cost and latency.
• Add retrieval, memory, and structured outputs where they improve accuracy.
• Instrument evaluations, guardrails, retries, and human review points so results stay trustworthy.
• Document how each workflow behaves, where it can fail, and what happens when it does.

Raise the team's leverage

• Automate internal work: data ingestion, quality checks, reporting, and routine client operations.
• Turn recurring product and support friction into durable engineering fixes.
• Help teammates use AI tooling well in their own day-to-day work.

• Features ship steadily and clients notice the product getting better.
• AI workflows run reliably in production, with clear accuracy expectations and sensible fallbacks.
• Manual internal work shrinks quarter over quarter.
• Nothing critical depends on undocumented knowledge living in one person's head.

• Several years of professional software engineering, with real ownership of production systems.
• Strong TypeScript or Python, plus comfort with SQL and modern web application architecture.
• Hands-on experience building with LLMs: tool calling, agent loops, retrieval, structured outputs, and evaluation.
• Judgment about when an agent is the right answer and when plain deterministic code is better.
• Clear written communication and comfort working independently in a small, fast-moving team.
• Familiarity with Hawaiʻi government, legislative work, campaigns, or civic organizations is a plus, not a requirement.

• A research position training or fine-tuning foundation models from scratch.
• A prompt-only role with no software engineering responsibility.
• A role where AI output ships to clients without evaluation or review.