Member of Technical Staff - Fraud

# Member of Technical Staff - Fraud **Location:** New York City or Mexico City, 80% in office ## About Nelo **Nelo** is a (profitable) leading consumer fintech and e-commerce platform in Mexico, with >$500MM in annualized GMV and >$70MM in annualized revenue. Our mission is to increase consumers' buying power in Latin America, and we're doing so by building an AI-native platform powered by credit. Nelo has raised over $40M of venture capital from investors including Homebrew, Two Sigma Ventures, and Susa Ventures. Nelo has additionally raised a $100M asset credit facility from Victory Park Capital. Our lean team includes experienced leaders from top technology companies such as Uber, Amazon, Rappi, and DiDi. We pride ourselves on our velocity, intellectual rigor, and efficiency. Nelo has offices in Mexico City and New York City. ## WHY THIS ROLE IS DIFFERENT In most companies, fraud is a cost center. You inherit a rules engine someone else built, you tune thresholds inside guardrails someone else set, and success means losses stayed flat. The mandate here is bigger. In your first three weeks you will run an independent assessment of Nelo's entire fraud surface and write the roadmap yourself. You will own detection, mitigation, and the operating function behind them, end to end, on a lending business moving more than $500MM in annualized GMV. The mandate is precision first: we would rather challenge almost nobody and be right than blanket good users with friction. You will build the fraud ops function as an agent-operated system from the start, and you will work directly with the founders. Whether a programmatic attack against Nelo is even possible six months from now depends on the controls you put in place. ## WHAT YOU'LL DO **Role mission:** Engineer a fraud detection and mitigation system that maximizes precision first, then recall. 1. **Fraud threat assessment.** Within three weeks of start date, produce an independent fraud risk assessment of Nelo. Where are our biggest weaknesses? What is on fire, or a greater risk than we previously believed? What roadmap follows from those findings? 2. **Step-up optimization.** Best-in-class log-in challenge rates are ultra-high precision and very low in relative terms. Within five weeks of start date, launch a test intended to get us to our challenge rate and precision goals. 3. **Acquisition fraud** (if your background is in financial services). Much of our fraud in dollar terms is first-party fraud at time of acquisition: a tough problem, and one rich in opportunity. Within one month of start date, launch an experiment that reduces first payment default rate by 3 percentage points for a segment, with the smallest possible impact to approval rate. 4. **Establish the fraud ops function.** Within three months of start date, build a system that generates repeatable intelligence from a manual fraud case review feedback loop, whether with AI agents, human agents, or both. 5. **Cap any currently uncapped fraud risk.** Within six months of start date, the risk of an unmitigated, programmatic fraud attack (acquisition, transaction, ATO) should be reduced to negligible because of controls you have put in place. This is the number one outcome for this role. ## WHY YOU SHOULD APPLY - **You have owned a fraud area before**: reviewing cases in your domain, deploying high-precision rules or models to production, and iterating until the fraudsters give up and go bother someone else. - **You run AI agents in a meaningful capacity today**, and you are building toward self-improving feedback loops for agentic systems. - **Your analysis stands on its own**. Nobody double-checks your work for technical correctness, and you understand every line of SQL your agents write. - **You step through every screen a user will see before a control ships**. You can intuit what good and bad product experiences look like, for good users and for fraudsters. - **You can take charge during an incident**, tell people exa…