Founding Full-Stack Engineer

Specter; New York (NY); On-site; USD 140,000 - 210,000; Full time

Job summary

Specter is building the surfaces where finance teams and AI agents collaborate, and you’ll ship across the full stack to ensure trust and accuracy in the data that powers decision making.

You’ll own features end to end, interact with customers, and establish high quality standards across the team, delivering reliable, calm user interfaces with thoughtful defaults.

Qualifications

• Strong React and TypeScript on the frontend, strong Python on the backend, real PostgreSQL comfort.
• Independent judgment: turn vague customer problems into working features.
• High bar on quality: fix edge cases like broken empty states and layout shifts.
• Nice to have: experience building tools for non-technical professionals or a design sense.

Responsibilities

• Own features end to end: design, frontend, API, schema, deploy and customer conversations.
• Build surfaces where humans supervise agents: live execution, reviews, audits, and editable know-how interfaces.
• Make the product feel calm with good defaults and no unnecessary modals or noisy UX.
• Collaborate directly with customers to translate feedback into shippable work without prolonged clarification.
• Raise quality standards through code/design reviews and robust shipping practices.

Skills

React

TypeScript

Python

PostgreSQL

Job description

The role

Specter's product is the surface where finance teams decide whether to trust an AI agent with their books. That trust is built or lost in the details - how a review queue is laid out, how an agent narrates its reasoning, how fast a page loads when a controller is closing the month at 11pm, what happens when an agent gets something wrong and someone has to fix it. This role exists for engineers who care about those details and ship across the whole stack to get them right.

You'll work on the surfaces where humans and agents share work: live execution, review and approval, audit and history, the interfaces that show what the system knows about a given company and let users correct it. These are the parts of the product customers spend the most time in, and they're how Specter earns the right to do more work autonomously over time.

What you'll do

• Own features end to end. Design, frontend, API, schema, deploy, and the conversation with the customer who asks you to change it the next morning.
• Build the surfaces where humans supervise agents: live execution, review and approval, audit logs, and the interfaces that let users see and edit what the system understands about their business.
• Make the product feel calm. Accounting work is stressful; the tool shouldn't be. Good defaults, no surprise modals, no information for its own sake.
• Make the right technical and product calls without waiting for a spec. Specter doesn't have a PM layer between engineering and the customer, and we don't plan to. You'll talk to customers directly and translate what you hear into what ships.
• Raise the bar on quality across the team. Code reviews, design reviews, the standard of what's allowed to ship - you set those by example.

What we look for

• Strong React and TypeScript on the frontend, strong Python on the backend, real Postgres comfort. You can move between layers without slowing down.
• Real product instinct. You've shipped something users actually loved, and you can explain why it worked at the level of specific decisions, not vibes.
• Independent judgment. You can take a vague problem from a customer call and produce a working feature without three rounds of clarification.
• High bar on quality. You notice the broken empty state, the off-by-one in the timestamp, the layout shift on load. You fix them before anyone has to file a ticket.

Nice to have

• Built tools for non-technical professionals - finance, legal, ops, healthcare. Domains where the user's day job is not software.
• Design background or strong design sense. Bonus if you have opinions about typography and don't apologize for them.
• Experience at a small team where you owned both the customer relationship and the codebase.