Senior Product Manager, Underwriting Specialist, CoStar Debt Solutions

Location: Boston, in-office 

CoStar Debt Solutions is building a new platform for CRE underwriting — one where credit officers, data pipelines, and AI agents work against the same shared data model. Banks, sponsors, and asset managers query, write, and collaborate on the same underlying system. 

We’re hiring a Senior Product Manager, Underwriting Specialist to help us build a world-class underwriting platform — the workflows, defaults, and decisions that decide whether a credit officer or originator reaches for our product first thing in the morning. In year one, that means turning underwriting scenarios into product: capital stack construction, rent roll injection, peer benchmarking, portfolio surveillance, deal counterparty visibility. In year two and beyond, it means owning the underwriting surface as it expands across asset classes, customer types, and the agent workflows that increasingly run alongside the human ones. 

The work starts with hard-won underwriting judgment. It ends with that judgment encoded into a system thousands of underwriters run against. 

What you’ll own: 

• The underwriting workflows across the product surface — capital stack construction, rent roll ingestion, peer benchmarking, portfolio surveillance, deal counterparty visibility. These exist today as scenarios; the work is turning them into product that originators and credit officers reach for daily. 

• The default behaviors that decide whether the platform feels expert or generic — the stack heuristics the LLM proposes, the way variance gets sorted, what the system asks vs. assumes, when it surfaces uncertainty, what stays in front of a human. 

• Design-partner relationships at two or three banks and sponsors. You will be on the calls, in the deals, and the person they trust to translate what they need into what gets built. 

• The boundary between “the platform can model this” and “the workflow should expose this” — working directly with engineering and product on where the line gets drawn for each new capability. 

• The roadmap for the underwriting surface as it grows — new asset classes, new customer profiles, new agent-driven workflows that sit alongside the human ones. You decide what’s on it and in what order. 

Basic Qualifications: 

• Deep CRE underwriting domain expertise. 

• 8 or more years, ideally as a credit officer at a bank with a real loan book. Sponsor-side, life co, or debt fund backgrounds are interesting too if the depth is there. Domain is the highest bar for this role — it’s what we can’t substitute for. 

• A working model of how originators, credit officers, asset managers, and sponsors actually spend their days — not just the artifacts they produce, but the friction between them. 

• Bachelor’s degree from an accredited, not-for-profit, in person, university or college. 

• A track record of commitment to prior employers 

Preferred Qualifications: 

• Sharp judgment about what’s worth automating, what needs human-in-the-loop, and what should never leave a credit committee. 

• A bias for being specific. Generic underwriting opinions are not useful here. We need someone who can say “the LLM should not propose mezz above 12% in this market without flagging it” and be right about it. 

What helps a lot: 

• Tech fluency. You don’t write production code, but you read a schema, reason about software as a system, and can sit next to engineers and argue with the model when it’s wrong. If you’ve worked closely with product or engineering teams before, that shows up immediately in this role. 

• Product sensibility. An instinct for what belongs in a default, what belongs in a setting, what belongs in a warning, and what belongs nowhere at all. Underwriters who have shipped tools — internal or otherwise — tend to have it. 

• Comfort with LLM-driven workflows. You don’t need to have built with them. You do need to find the question of “what should the model decide vs. ask vs. flag” genuinely interesting. 

The growth path: 

The first year is product-heavy: real workflows, real design partners, real deals informing real shipping decisions. The second year shifts. As the underwriting surface expands and agent-driven workflows take on more of the routine work, the role grows from “shape these workflows” into “own the underwriting platform end-to-end” — defaults, asset classes, the line between what the model decides, what it asks, and what it escalates, the whole surface. The person who owns this role is the one whose judgment ends up shaping how an entire market underwrites. 

What’s unusual about this role: 

You will not be the underwriter of record on anything. You will be the person who decides what the platform knows and does on behalf of every underwriter who uses it. That is a different job — more leverage, more abstraction, and more responsibility for the second-order effects of the workflows you ship. People who want to keep doing deals will be unhappy here. People who want their judgment encoded into a system that thousands of other underwriters will run against will be in the right place.