Senior Full Stack Software Engineer

Eddi; United States; On-site; USD 140,000 - 180,000; Full time

Job summary

Eddi is seeking a senior full-stack engineer to own the TypeScript platform with a focus on AI-driven agents and live data analysis. You will contribute across the stack, including conversational analysis over live data, SQL planning, document extraction, and durable workflows that power grant management for schools and districts.

You'll tackle multi-dataset reasoning, evaluation harness design, and robust UI work where humans steer agent output.

Qualifications

• 7+ years building production software with TypeScript.
• Experience shipping LLM-powered features to production (agents, tool use, structured output).
• Solid backend and distributed-systems fundamentals (Node.js, Postgres; retries, idempotency, partial failure).
• Temporal experience is a big bonus; durable workflow orchestration core to the platform.

Skills

TypeScript

LLM integration

Node.js

Postgres

Distributed systems

Temporal

React

Full-stack ownership

Ambiguity handling

Tools

Node.js

Postgres

Temporal

Job description

Eddi is building the AI-powered grant management and financial management platform for schools, districts, and mission-driven organizations, teams that run on grant funding but manage it with spreadsheets and legacy systems. Our AI doesn't just store their data; it analyzes it, answering questions over live financial data in plain language, surfacing insights across programs and budgets, and generating the funder reports that used to take administrators weeks to assemble. Every hour we give back to a grants administrator goes toward students and communities instead of paperwork.

We're a small, early-stage team. The product is live, customers depend on it daily, and this is the stage where one engineer's decisions become the architecture everyone builds on.

The Role

You'll be a senior full-stack engineer across our TypeScript platform, with a center of gravity in AI and agents. Agents are a core product surface at Eddi, not a feature bolted on. We run many agents built on an AI SDK, covering conversational analysis over customers' live data, SQL query planning against runtime-defined datasets, document and grant extraction from messy PDFs, report drafting, and grant matching. Long-running agent work runs on durable workflows.

Problems on the table right now: making a conversational agent reason correctly across joined datasets it has never seen before, designing evals that catch agent regressions before customers do, keeping multi-step AI workflows correct when any step can fail or take minutes, and building UI where humans review, steer, and trust agent output.

What We're Looking For

• 7+ years building production software with deep, current TypeScript fluency. We run strict TS everywhere and favor a functional style.
• Experience shipping LLM-powered features to production — agents, tool use, structured output, and evaluating them. If you've built an eval harness, or been burned badly enough to want one, we want to hear about it.
• Solid backend and distributed-systems fundamentals : Node.js services, Postgres, and designing for retries, idempotency, and partial failure.
• Temporal experience is a big bonus — durable workflow orchestration is the backbone of our platform; half our services are Temporal workers.
• Real full-stack range. Modern React should be a place you're productive, not a place you visit reluctantly. Deep design-system expertise isn't required; owning the whole vertical is.
• Comfort with early-stage ambiguity. You fill gaps with judgment, not tickets.

Who You Are

• Extreme ownership is your default. You run toward broken things, and a task is done when it's verified working in production, not when the PR merges.
• You're driven. You hold a high bar because you want to, and you root-cause instead of patching symptoms.
• You want your work to matter. What you build here determines whether a school district actually gets and keeps the funding it's owed.

Who we are

Small team, high trust, low ceremony. We plan before we build, verify before we call anything done — we ship it when we'd stake our name on it, and keep blast radius small. We build agents — and we work alongside them aggressively in our own workflow too. Everyone is close to customers, and your work is visible in the product within days, not quarters.

• Salary: $140k-$180k depending on experience