ML Systems & Production AI

Real-time ML feature pipelines, model deployment & governance, staged rollouts, A/B testing, low-latency ML inference

Distributed Systems & Infrastructure

AWS, Kubernetes, Kafka/Kinesis, Snowflake, Apache Iceberg, Trino, Druid, Spring Boot

Languages

Kotlin, Java, Python, SQL, JavaScript

• Lead SoFi's Fraud & AML engineering org, spanning fraud, data, and AML engineering teams across US and European time zones. Manage both engineers and the engineering manager who leads AML Engineering.
• Led fraud and AML engineering for all 8 of the CEO's company priorities in the past year, including net-new products like crypto withdrawals, SoFi Pay, and business banking, where fraud controls were built from scratch. Launched crypto withdrawals to 493K members with fraud losses 80% below the limit set for launch.
• Created a new model for how fraud engineering supports launches: one fraud engineer embeds with the product team as a forward-deployed engineer, and the product team writes and owns the integration code. It's now the default for new launches and has tripled the number of concurrent initiatives fraud engineering can support.
• Hired engineers across IC levels, promoted engineers to Senior, Staff, and Senior Staff, and handled coaching and performance management. Developed a Staff engineer into a manager.
• Raised outbound instant-payment volume 23% and cut blocked transfers 89% by replacing fixed transfer caps with risk-based decisions, tested on 2.1M members with no rise in fraud losses.
• Rebuilt transfer limits and fraud controls across 5 payment rails for SoFi's core banking migration. Members now get same-day access to 42% of the money they move in from other banks, up from 15%.
• Own SoFi's fraud and AML decisioning platform, returning real-time decisions in under 20ms p95 for banking, lending, and investing products. Built the fraud data lakehouse its models and rules are trained and tested on (260M+ transactions, 28K labeled fraud cases).
• Consolidated new and legacy AML pipelines onto the risk platform. Payment products now get fraud and AML controls through one integration, and SoFi's first in-house AML risk model scores 11.5M customers a month in place of a vendor model.

• Core contributor and tech lead for SoFi's 2nd-generation financial services platform. Architected and led development of the product-agnostic onboarding platform, a concurrent and resilient framework behind member onboarding across SoFi products.
• Authored the FRN (Financial Resource Name) spec, a URN-style identifier standard adopted across SoFi's 100+ microservices for resource identification and authorization.
• Team lead and backend engineer on the SoFi Money team, building Kotlin, Java, and Spring Boot microservices.
• Tech lead on large member-facing launches, including Samsung Money by SoFi and the Personal Loan + Money bundle, both aimed at growing direct-deposit-enabled accounts, a top company KPI. Featured in a blog post by builtin.com.
• Designed and built the webhooks API SoFi uses to communicate with Samsung: 100M+ messages sent with zero dropped.
• Designed and implemented the org-wide data pipeline core infrastructure on AWS Kinesis, S3, and Snowflake.
• Started on the Time To Fund team, using automation to reduce friction between signing and funding.

• OTC Exchange Network (“OTCXN”) was a capital markets infrastructure company using proprietary blockchain technology.
• Full stack development with Java micro-service backend and React.js frontend

• Consult with departments within Analytical Mechanics Associates at NASA to provide both local and online software solutions.
• Use Java, VB, Microsoft SQL to code CRUD application solutions to assist in the research of thermo-protective materials to be used in NASA reentry vehicles.
• Designed and developed relational database application for documenting the decomposition products of aerospace materials being tested under atmospheric reentry conditions.
• Custom UI design built with JavaScript, CSS and HTML enabled NASA engineers with no DB training to customize their own queries, generating reports and data visualizations on demand.
• Presented to NASA administrative staff software capabilities and results of the investigation.
• Check out a PDF of my poster presentation HERE.