Staff AI Product Engineer

Compensation: $160k – $230k • 0.5% – 1.5%

The role

Deeply is building a guided workspace that helps couples understand what is happening in their relationship, decide what needs attention, and change recurring patterns.

We are looking for a staff engineer who can take an ambiguous customer problem from first principles through product definition, technical design, implementation, and real-user validation.

This is not a ticket-execution role. You will be expected to help determine what should be built, make the product concept coherent, ship it rapidly using AI, and learn from how couples actually use it.

Your primary job

Turn fuzzy customer problems into simple, valuable, production-quality product experiences.

A successful project looks like:

1. Understand the customer problem and relevant domain.
2. Identify the specific job the product must accomplish.
3. Model the underlying concepts, states, and workflows.
4. Evaluate alternatives and recommend a direction.
5. Define the smallest valuable, testable version.
6. Prototype and implement it rapidly.
7. Get the team and customers to use it.
8. Analyze what happened and improve the model and product.

The founder should provide strategic and customer context—not have to perform every step of this decomposition for you.

What you will own

• Major product areas from concept through production
• AI-guided conversations and structured outputs
• Durable customer artifacts, such as shared understandings and protocols
• Multiplayer workflows involving two partners at different stages
• Product state, permissions, provenance, and trust boundaries
• LLM context, memory, structured models, and evaluation
• Full-stack implementation across UI, APIs, database, and background workflows
• Instrumentation, testing, debugging, and production reliability
• Internal tools that make AI-assisted development safer and faster
• Closing the loop with internal testers and real couples

What exceptional performance looks like

• You bring recommendations rather than unresolved ingredients.
• You make complicated concepts feel simple to customers.
• You know when to prototype quickly and when to build durable infrastructure.
• You surface product and technical risks before they become blockers.
• You ship meaningful, customer-testable increments every week.
• You drive testing and follow-through without being reminded.
• You use AI to produce extraordinary throughput without creating incoherent code.
• Your presence materially reduces the founder’s product and technical cognitive load.
• The product becomes more coherent, reliable, and valuable because you own it.

AI-native engineering expectations

We expect AI to be part of your normal operating system, not merely an autocomplete tool.

You should be able to:

• Give coding agents sufficient architectural and product context
• Decompose projects into parallelizable workstreams
• Use AI to explore unfamiliar systems and compare approaches
• Generate and review implementations, migrations, tests, and documentation
• Detect incorrect assumptions and low-quality generated code
• Build evaluation and verification loops around AI output
• Maintain consistency across a rapidly changing codebase
• Compress work that traditionally took weeks into days

We measure AI capability through resulting speed, quality, and judgment—not the number of tools or prompts used.

What we are looking for

• Evidence of building and shipping meaningful zero-to-one products
• Strong product judgment alongside strong engineering judgment
• Excellent full-stack engineering ability
• Experience with modern web applications, relational data, APIs, and asynchronous workflows
• Experience building production LLM features, including structured output and evaluations
• Ability to reason about complex state, permissions, identity, and data integrity
• Strong written communication and conceptual modeling
• Comfort working directly with customers and imperfect qualitative evidence
• High agency, intellectual curiosity, and willingness to develop domain expertise
• Ability to disagree clearly, explain tradeoffs, and then drive decisions to closure

Big-company pedigree is neither necessary nor sufficient. We care about what you personally understood, decided, built, and caused to happen.

This role is probably not right for you if

• You need detailed tickets before beginning work.
• You consider a merged PR to be the completion of the job.
• You prefer implementation to product reasoning.
• You wait for others to resolve ambiguity.
• You produce many parallel initiatives without prioritizing the customer outcome.
• You use AI primarily to generate code faster.
• You build elaborate systems before validating whether the product matters.
• You want a narrowly bounded engineering role.