Senior Full-Stack AI Engineer

RealQuant · Remote for now, NYC preferred; hybrid in NYC once our office opens · Full-time

About RealQuant

RealQuant builds AI infrastructure for institutional commercial real estate. We deploy governed data, deterministic models and cited, guardrailed AI inside each client's own Azure tenant, for private equity firms, owner-operators and lenders who want to grow AUM without adding headcount. AI never does the math.

Our founder spent 12+ years on the institutional buy side, with 150+ transactions and $11B+ in deal volume at platforms backed by Blackstone, Ares and Angelo Gordon. We are bootstrapped and growing.

The role

This is our first senior engineering hire and a founding-engineer seat. You will set technical direction for RealQuant Labs, where we design, build and run AI infrastructure for our clients, and for the labs.realquant.ai workflow platform and document parsers that power it. We expect you to propose what we build next, not just build what is asked.

The core of the job is turning client work into product:

• Build the library. Turn repeatable client workflows into modular, reusable components: universal schemas and data models, prebuilt ETL pipelines, MCP servers and front-end shells.
• Deploy it. Lead client projects that assemble those components into production systems in each client's Azure subscription.
• Productize it. Package the best of it as off-the-shelf APIs and SaaS, starting with our document parsers and Excel add-in.
• Embed when it counts. Take select engagements working directly inside a client's team.
• Build the team. Help hire and lead engineers in Latin America and Europe, and grow into a lead role as we add data engineers, data scientists and junior staff.

What you will build

• Deal workflows on labs.realquant.ai. Broker listing sourcing, mailbox intake into triage and extraction, buy-box screening and ranking, a deal pipeline synced to Monday.com, lease abstraction, comps and market research, portfolio reporting, a text-to-SQL analyst chat, and generated Excel models, decks and PDF reports.
• Data foundations. PostgreSQL and Databricks in client tenants, a CRE data model (properties, deals, leases, comps, funds, debt), and connectors to property systems, data vendors and public data.
• Document parsers. Extraction for offering memoranda, rent rolls and T-12 P&Ls, with a citation and confidence score on every field and human review before anything reaches a client.
• Accuracy you can prove. Underwriting math in auditable code, deterministic checks (totals foot, NOI ties), cross-document reconciliation, and evals that gate every release.
• The AI layer. MCP servers, Claude skills and plugins that let analysts query their own data in plain English, with guardrails, citations and an audit trail.
• Web portals. React (Vite and Next.js) apps for data review, QA and map-based comp search, portfolio and debt dashboards, and Entra ID sign-in.
• Expert data and evals (secondary). Support our platform where CRE finance experts author underwriting tasks with gold answers, rubrics and graders, packaged as evaluation suites and RL environments (Inspect, Prime Verifiers) for testing frontier models.

What success looks like

• 30 days: you have shipped a fix to a live client deployment and traced a deal end to end through labs.realquant.ai.
• 90 days: you own a client build and have turned at least one piece of it into a reusable library component.
• 6 months: you lead deployments end to end, set our technical roadmap with the founder, and are helping hire the next engineers.

The stack

• Languages: TypeScript (strict, Node 20+) and Python
• Front end: React (Vite and Next.js), Tailwind, shadcn/Radix, TanStack, AG Grid, Mapbox
• Back end: Express and Next.js route handlers in TypeScript, FastAPI in Python, REST / OpenAPI, SSE, Postgres-backed job queues
• Data: PostgreSQL on Azure Flexible Server (row-level security, PostGIS, pgvector), raw SQL with hand-written migrations, Databricks (Delta, Unity Catalog) for lakehouse clients, Power BI semantic models
• AI: Anthropic Claude (direct and through Azure AI Foundry), Azure OpenAI, Foundry Agent Service, Azure Document Intelligence, Azure AI Search, Model Context Protocol servers, Claude skills and plugins, LangGraph
• Evals and RL: deterministic graders, rubric-based LLM judges, golden-replay regression suites, Inspect AI, Prime Verifiers, Label Studio
• Microsoft 365: Entra ID, Microsoft Graph (mailbox intake, Teams, SharePoint), Excel (exceljs, Office add-ins, custom functions)
• Infrastructure: Azure Container Apps and jobs, Blob Storage, Key Vault, private networking, Bicep, GitHub Actions, Docker
• Quality: CI eval gates, Playwright, Vitest, pytest, confidence calibration

Must-have

• A track record of building and shipping something from scratch on your own initiative.
• Bachelor's degree or higher in computer science, engineering, math or a related field from a top-tier university.
• 5+ years of full-stack engineering, shipping production systems end to end.
• Strong TypeScript, React, Python and SQL.
• Production experience on Microsoft Azure.
• You have built MCP servers or agent tools that connect LLMs to real data and systems.
• You have shipped LLM features to production and built the evals that prove they work.
• You work AI-native: Claude Code or similar agentic coding tools are part of your daily workflow.
• You have turned messy PDFs and spreadsheets into clean, structured data.
• You communicate clearly with clients and executives.
• Based in the US and authorized to work here. NYC area strongly preferred.

Strong-to-have

• Experience at a bank, asset manager, investment firm, or proptech or fintech company.
• Evals, labeling pipelines, graders or RL environments (Inspect, Verifiers, RLVR-style reward functions), confidence calibration, or familiarity with RL post-training methods such as GRPO and DPO.
• Databricks (Delta, Unity Catalog).
• Microsoft Graph, Power BI or Excel add-in development.
• Hiring or leading engineers, including distributed teams.
• OCR and table reconstruction, or geospatial data (PostGIS, Mapbox).
• Real estate, private equity or financial analysis knowledge. Learnable; the founder will teach you.

Who thrives here

• You have built something from zero on your own initiative: a company, a product or a serious side project.
• You think like an owner and a founder, about customers, product and the business as well as code.
• You see where AI is going and want to build the category, not a feature.
• You ship without being asked and bring the plan with you.
• You are comfortable with ambiguity and a small team.
• You would rather prove accuracy with an eval than argue about it.
• You hold buy-side standards: in financial software, a misplaced decimal is a real-money error.
• When a critical client issue lands on a weekend, you fix it.

How we work

• Full-time salaried (W-2), minimum 40 hours per week, US Eastern hours.
• Remote for now. We are opening a New York office and expect the role to become hybrid in NYC, so candidates in or near New York are strongly preferred.
• Occasional travel to client sites, mostly in the New York area.
• Pay: $125,000 to $150,000 base salary, plus an annual performance bonus. Final offer depends on experience.
• Benefits: Health, dental and vision insurance, and paid time off.
• Client work runs on a company-provisioned Microsoft Cloud PC with time-tracking software for client billing. Activity on company systems is monitored.
• You report directly to the founder.
• Our process: a call with the founder, a technical deep dive on something you have built, a working session on a real problem, and references.

RealQuant is an equal opportunity employer. We do not discriminate on the basis of race, color, religion, sex, sexual orientation, gender identity, national origin, age, disability, veteran status, or any other protected characteristic. We are unable to sponsor visas for this role.