Forward Deploy Engineer

Shiboleth.ai (YC W24)

San Francisco (CA)

On-site

USD 120,000 - 150,000

Full time

14 days+

Get more replies from employers

Send a job-specific resume in minutes.

Benefits offered by this job

Comprehensive healthcare

Unlimited PTO

Paid relocation and temporary housing

Latest technology and office snacks

Job summary

A pioneering AI compliance firm in San Francisco seeks a candidate to deploy compliance solutions at banks and fintechs. You will ingest data, configure rules, and deliver accurate audit findings. The ideal candidate has strong SQL skills, a background in data pipeline architecture, and an ability to communicate effectively with customers. Benefits include comprehensive healthcare, unlimited PTO, and flexible work arrangements.

Qualifications

Experience deploying compliance solutions for banks and fintechs.
Strong problem-solving skills with a focus on customer needs.
Ability to write and debug SQL rules and data ingestion pipelines.

Responsibilities

Deploy Shiboleth compliance solutions at customer sites.
Build and modify compliance rules based on customer requirements.
Identify patterns and iterate on product features based on customer feedback.

Skills

SQL auditing

Data pipeline architecture

Customer communication

Debugging skills

Tools

DuckDB

SFTP

APIs

Job description

Overview

tl;dr: We’re automating compliance testing and monitoring for banks and fintechs with AI — replacing manual audit workbooks with intelligent, continuous oversight. Our FDEs are the bridge between our platform and our customers’ messy, real-world compliance problems. You’ll deploy Shiboleth at partner banks and fintechs, build custom rules and integrations, and feed learnings back into the product. If you love solving hard problems at the intersection of engineering and customer impact, we should talk.

Before a fintech can issue a loan or process a dispute, its partner bank has to prove the whole operation is compliant — Reg E, TILA, FCRA, SCRA, UDAAP, BSA and dozens more. Today, that means consultants spending weeks manually sampling files, writing findings in spreadsheets, and hoping nothing slips through. Shiboleth (YC W24) builds the AI-powered compliance testing and monitoring platform for the Banking-as-a-Service ecosystem. We test every file, every transaction, every notice — not a 5% sample. Our customers include the banks and fintechs that power modern lending, payments, and deposits.

What You’ll Do

Deploy Shiboleth at customer sites. You’ll work directly with partner banks and fintechs — ingesting their data, configuring compliance rules, building integrations with their systems (SFTP pipelines, APIs like Peach Finance, email platforms), and making sure the platform delivers real audit findings from day one.

Build custom compliance rules. Every customer has different products, different data schemas, different edge cases. You’ll write and tune SQL-based audit rules in DuckDB, debug false positives, and validate findings against regulatory requirements like Reg E provisional credit timelines or TILA disclosure accuracy.

Close the feedback loop. You’ll be the closest person to our customers’ pain. You’ll identify patterns across deployments, spec product improvements, and ship features back into the platform — from better dashboards to smarter LLM evaluation pipelines.

Own outcomes, not tickets. When a bank needs population-level Reg E testing across 14,000+ disputes by next week, you figure out how to make it happen. You’ll architect solutions, write code, talk to customers, and deliver results — often in the same day.

Technical Challenges

Population-Level Compliance Testing
Traditional auditors sample 5% of files and extrapolate. We test 100% — every dispute, every loan, every notice. This means building data pipelines that ingest messy, inconsistent customer data (CSVs, SFTP drops, API pulls) and running thousands of audit rules against millions of records with sub-hour turnaround.

LLM-as-Judge Evaluation
Some compliance questions can’t be reduced to SQL — “Was this adverse action notice clear and conspicuous?” requires judgment. We’re building LLM evaluation frameworks that simulate human analyst reasoning, with calibration runs to measure consistency and accuracy across regulatory standards.

Data Integration Across a Fragmented Ecosystem
Every bank and fintech stores data differently. You’ll build and maintain integrations across loan management systems, dispute platforms, document stores, and email providers.

Compliance Rules That Actually Work
Writing a rule is easy. Writing one that catches real violations without drowning the customer in false positives is hard. You’ll iterate on rule logic across Reg E, TILA, FCRA, SCRA, UDAAP, and more — balancing precision and recall against regulatory expectations.

What We Look For

You solve problems end-to-end. You’ll architect data pipelines, write audit rules, debug SQL edge cases, build integrations, and present findings to bank compliance officers — often in the same week.

You have a high bar for quality. In compliance, “mostly working” means audit findings that are wrong. You obsess over edge cases, build robust error handling, and ship systems that institutions trust with their regulatory obligations.

You’re energized by customer problems. You’d rather spend an hour on a call understanding how a bank processes Reg E disputes than a week building something nobody asked for. You work backwards from the customer’s pain.

You communicate with clarity. You need to articulate tradeoffs to both engineers and compliance officers. You can explain a false positive root cause to a non-technical bank examiner and spec a data pipeline fix in the same conversation.

You ship fast and iterate. We move fast with minimal process. You’re comfortable deploying something that works today and improving it tomorrow, without sacrificing the quality bar that regulated customers demand.

You hate meetings. You would much rather focus your time on building, being productive, and shipping.

About the Interview