Founding Engineer (Stackpoint Portfolio Company)

Job Description

Job Description

Stackpoint is a venture studio. Each year we build and fund 3–4 new vertical AI companies that are transforming legacy industries. A few months after each one launches, we hire its Founding Engineer.

Because of this model, we're continuously meeting engineers who may be a fit for current or upcoming roles. If you want real ownership, real impact, and to build alongside exceptional builders, we'd love to connect, even if the timing isn't perfect yet. Getting to know you early helps us put the right opportunities in front of you when they open.

The Role

You'll join an early-stage company with customers and a product already in use.

As one of the company's first engineers, you'll work with the founding CEO, product leadership, and a small engineering team as it forms. Stackpoint's studio team builds the first version of each platform with the founding team, design partners, and early customers. Your job is to take that platform from its first customers to many, and to build the new workflows and products the company needs next.

The work is both, constantly: new products and workflows from a blank page, and scaling the ones customers already depend on. The products are AI-native. Agents do work that people in the industry do by hand today, human operators review and correct AI output, and B2B customers use the product directly. There is no separate DevOps, SRE, QA, or data team. The engineering team owns all of it together.

What You'll Drive

1. Build new workflows and products from scratch

• Take new workflows, services, and customer segments from idea to production, working directly with product, operations, and customers.
• Make the "build to learn" versus "build to scale" call explicitly, and throw away a prototype when the team learns something important.

2. Scale what's working

• Extend a working platform built with the studio team, and take it from first customers to many.
• Find the bottlenecks before customers do: performance, reliability, cost, and failure modes.
• Build foundations so onboarding the next customer or market is configuration, not a project.
• Ship, deploy, and monitor what you build, including CI/CD, logging, and alerting.
• Maintain and extend the internal tools and agents the team relies on to operate.

3. Ship agentic AI workflows that hold up in production

• Build extraction, classification, and retrieval over messy, unstructured industry data: documents, forms, scans, and records.
• Orchestrate multi-step agents with tool use, structured outputs, and deterministic guardrails.
• Set the confidence thresholds that decide what goes to a human reviewer.
• Make model selection and cost/latency tradeoffs with data.

4. Integrate with systems that weren't built to be integrated with

• Connect to the legacy software, web portals, and email and file-based workflows that high-barrier industries still run on, including agent-driven browser automation where there's no API.
• Build reusable integration patterns where volume justifies it, and optimize the paths that carry the most volume.

5. Build the human-facing product

• The internal operations console: the human-in-the-loop surface where operations and servicing staff review, correct, and approve AI output quickly. Their corrections feed back into evaluation.
• The client-facing B2B experience: how customers submit work, track status, and act on results.
• Handle PII and sensitive financial data correctly across both, with access control, audit trails, and tenant isolation.

How We Work

• Build with modern agent architectures and the latest AI techniques, using data and evals to drive decisions.
• Own problems end to end, from scoping to shipping to monitoring in production.
• Choose the 80% solution that ships today over the perfect one that ships next month.
• Build evaluation before claims: instrument accuracy, regression, and confidence before calling something done.
• Work in the open: async-first updates, decision logs, and visible progress in shared tools.

What You Bring

Building and scaling early products. You've taken products from a blank page to launch, and you've carried a product past its first customers and handled what came with that: data growth, new tenants, reliability expectations, a codebase more people touch.

Production agentic AI. You've shipped LLM-powered workflows that real users depended on, and you can explain how you measured whether they worked: structured outputs, retrieval, evals, prompt design, and model tradeoffs.

A generalizer's instinct. When you see twenty similar problems, you build the abstraction. When one path carries most of the volume, you optimize it.

Full-stack range, including infrastructure. You work across a Python or Node backend, a TypeScript/React/Next.js frontend, relational and vector databases, and AWS or GCP. You can take something from local to production and keep it healthy, and you treat PII and sensitive data as a design constraint from the start.

Product sense for operational users. You've built tools for people who use them all day, and you care about the review screen as much as the model behind it. You'll spend time with operators and customers in discovery to see where the time goes.

AI-native in how you work. Coding agents and AI tools are core to your workflow, and you stay current on what actually speeds you up.

Experience

• 5+ years of software engineering, with meaningful time in startup or high-ownership environments.
• At least one product shipped from blank page to production, plus experience operating a system through a growth phase.
• Production LLM integrations: structured extraction, RAG, and agentic workflows.
• Data pipelines that take unstructured documents (PDFs, scans, web content) into SQL and vector stores.
• Hands-on ownership of cloud infrastructure and CI/CD.

Bonus Points

• Early engineer at a company that grew past its first handful of engineers.
• Browser automation or RPA running at production scale.
• Document AI, OCR, or multimodal models in production.
• Human-in-the-loop or operator tooling for AI systems.
• Multi-tenant B2B SaaS.
• Security and compliance: access control, secrets management, audit logging, data privacy, and SOC 2 readiness.
• Experience in real estate, construction, insurance, financial services, or other regulated industries.
• Prior venture studio or product-discovery consulting experience.