CRE Credit Specialist

Prysm builds AI agents that do the work of a commercial real estate credit team.

A broker package goes in; an institutional-grade screening deck comes out in under an hour, with every fact one click from its source document.

The workflow tools that exist today record what a credit team decided.

Prysm does the work.

The platform is in production and founding clients are onboarding.

The founding team pairs deep credit experience with senior engineering: a CEO who spent twenty years in European real estate credit, most recently as Partner and Head of European Real Estate Credit at Ares; a CTO and former Deloitte partner who led financial modelling across $9 billion of transactions and has spent the last three years entirely in applied AI; and a Head of Product with fifteen years in commercial real estate, formerly an Executive Director in Goldman Sachs' Special Situations Group.

The role

This is a new, full time position at the centre of the company, based in London with hybrid working. You will work directly with the Head of Product and the founders.

Prysm's defensibility is not the AI models, which it deliberately rents in the same way other leading vertical AI companies do.

It is the configuration library: twenty years of market knowledge written down in a form machines can execute.

Asset class taxonomies, geography conventions, broker methodology maps, screening logic, comparable evidence standards, output formats and evaluation criteria.

Your job is to grow, maintain and quality-control that library, and to make the agents measurably better at credit work every week.

This is a product role for a credit professional, not a software engineering role.

You will not write application code.

You will write precise English, the definitions, rules and instructions that steer the agents, and you will test what you write against real transactions until the output meets the standard an investment committee expects.

The tools that turn that writing into the platform's working files are simple to pick up, and the AI itself does most of the formatting; we will teach you the workflow in your first week.

What you will do

• Encode domain knowledge.

Translate CRE credit expertise into the precise definitions and rules that drive the agents: asset class definitions, market data conventions (which broker's vacancy series covers which universe, what "prime yield" means in each context, when two data points must never be averaged), screening criteria and comparable evidence standards.

• Work with AI models daily.

Write, test and refine the prompts that instruct the agents; run structured evaluations of their output; diagnose why an output missed the mark and fix what caused it.

• Quality assurance.

Review agent-produced screening decks the way an investment committee member would: check numbers against source documents, challenge the market narrative, and catch what a credit committee would catch.

• Client configuration.

Own the setup of new clients on the platform: screener definitions, house formats and market configurations.

Onboarding is configuration driven and takes days to weeks rather than months; you are the person who makes that true.

• Product definition.

Work with the Head of Product on the roadmap: translate the credit workflow into specifications engineering can build, and feed what you learn from clients and quality assurance back into the product.

Who we are looking for

• 3 to 5 years in commercial real estate credit or a directly adjacent seat: CRE lending, real estate debt funds, debt advisory, or real estate focused investment banking or credit ratings work.

You have underwritten transactions and written or reviewed screening papers and investment committee memos.

• Strong Excel modelling. You have built and stress-tested debt cashflow models and can take apart someone else's: rent rolls, debt sizing, covenant tests, sensitivities.
• A scientific mindset.

You are precise about definitions, alert to how data is constructed (survey universe, measurement basis, methodology changes between sources) and unwilling to compare numbers that are not comparable.

Given the choice, you write the exact rule rather than the vague one.

• Hands-on engagement with AI.

You already use tools like Chat GPT or Claude seriously in your own work and want to go much deeper.

No technical background is required: if you can write a precise credit memo, the rest can be taught in days.

• Excellent written English. Much of the job is writing things down exactly.
• Exposure to several asset classes (logistics, offices, living sectors, hotels) or several European markets
• Working familiarity with market data sources: broker research, Co Star, MSCI/RCA, Green Street or similar
• Any prior experience of prompt engineering, evaluating LLM output, or taxonomy and data-structure work

What we offer

• Competitive salary plus meaningful equity options
• A seat at the ground floor of a revenue-generating AI company built for your own market
• Daily work alongside a founding team with senior credentials in credit, product and engineering
• A role that barely exists elsewhere yet: the person who teaches the AI the market.

The skills this seat builds (domain encoding, model evaluation, agent quality assurance) are the skills the next decade of financial services will run on

• London based, hybrid working
• #J-18808-Ljbffr