Leadership & Cross-Functional Collaboration

• Acted as tech lead for KYC/KYB data ingestion and processing systems; partnered cross-functionally with Compliance and Risk stakeholders to meet aggressive timelines and deliver clarity across domains

• Drove alignment across engineering to reduce scope creep, saving an estimated sprint of development effort

• Shadowed Ops team workflows and integrated feedback into the Case Management platform to improve usability and efficiency

• Navigated the transition from legacy systems (Poetry, Zookeeper, CircleCI) to modern stacks (Nix, ArgoCD, GitHub Actions) while maintaining velocity on key deliverables

Infrastructure & Platform Engineering

• Migrated CI/CD workflows from CircleCI to GitHub Actions, reducing overhead by consolidating credential setup and replacing legacy codegen with a shared proto library

• Replaced outdated Lambda + API Gateway logging flow by designing and deploying an Envoy-based logging filter with traffic routing via internal load balancer

Service Development & System Design

• Developed and deployed the first endpoint for the Case Management service in under four weeks, integrating with Unit21 via request/response lifecycle handling

• Built foundational infrastructure using SQLAlchemy models, Alembic migrations, protobuf contracts, and endpoint definitions

• Led iterative design reviews for proxy service consolidation, delivering highly visual and well-structured architecture diagrams to facilitate team understanding during periods of limited stakeholder availability

Developer Experience & Documentation

• Updated Datadog monitors, runbooks, and retro templates to ensure accuracy, reproducibility, and ease of use

• Introduced and maintained story-based local development environments to reduce ramp-up time for engineers and minimize team-wide context switching

• Regularly maintained on-call playbooks and incident response logs, addressing DMS failures and Airbyte/Airflow data pipeline issues

• Spearheaded the end-to-end development of a customizable notifications system, enabling automated email alerts based on user-defined triggers—resulting in improved customer engagement and faster visibility into key metrics.

• Engineered a flexible alerting framework to support projected stockouts, rate-of-sale anomalies, and additional critical alerts.

• Designed and implemented a Knock workflow to batch and trigger daily email notifications when alert conditions are met.

• Backfilled and enriched alert staging data by linking merchant country IDs to customer subscriptions, ensuring accurate targeting.

• Led the design and launch of account relationships functionality, empowering parent accounts to manage child accounts and simplifying administration for enterprise clients.

• Built internal-facing services and APIs enabling agency clients to efficiently manage large account hierarchies.

• Rolled out a dedicated internal API instance for account endpoints, reducing load on the public API and significantly improving system performance.

• Collaborated with DevOps to provision infrastructure using Ingress Terraform, configure Convox, and integrate with Argo Workflows.

• Coordinated with internal stakeholders to reroute traffic from the public API to the new internal route, enhancing stability and performance.

• Instrumented Datadog dashboards to monitor usage and ensure smooth adoption of the new internal API path.

Financial Reporting Automation & Accounting Tools

• Led design and implementation of an API-driven worker to automate month-end report generation for the accounting team, eliminating hours of idle time spent waiting on 10+ reports.

• Built a message-driven worker leveraging internal queues with retry mechanisms to handle failures and delayed report generation reliably.

• Built a bulk expense upload API for insurance and property taxes, eliminating repetitive manual entry and significantly improving accounting team efficiency.

• Enhanced reporting accuracy by supporting actual-value fields and integrating them into the servicing tape report.

• Created Stripe fee reports for a cash flow management system to track the full payment lifecycle across platform and operating accounts.

• Automated calculation of internal fund transfers, reducing manual accounting overhead.

• Added user-facing UI option to download fee reports by line item for greater transparency and flexibility.

Transaction Categorization & Payments

• Revamped transaction categorization UI to support multi-property selection and specific amount allocation.

• Validated allocations against total transaction amounts and enabled uncategorization workflows with confirmation dialogs.

• Enhanced GraphQL queries and mutations to support advanced categorization logic.

• Developed payloads for bank reversal events and modified backend validation logic to account for these scenarios within the payments framework.

UI Modernization & Reporting Enhancements

• Migrated legacy JSX-based modal and form components to modern TSX functional components; added robust monetary validation logic.

• Updated and aligned UI with internal Core design system for consistency and maintainability.

• Improved rent run reporting by capturing card transactions and refactoring core helpers.

• Built runner scripts to test internal reports, improving confidence in reporting outputs and enabling repeatable validations.

Operationalizing Computer Vision for Real Estate – Image Feature Extraction Pipeline

Led the end-to-end engineering effort to operationalize Zillow’s first Computer Vision model, transforming raw home photos into high-quality labels for kitchen/bathroom finishes and appliance colors—key inputs for pricing analysts and AVMs (Automated Valuation Models).

• Enhanced pricing precision for Zillow Offers by increasing both automated and human-assisted valuation accuracy.

• Labeled ~2 million properties with >85% agreement to analyst-reviewed ground truth, significantly scaling Zillow’s home intelligence.

• Boosted analyst efficiency by automating visual assessments of home interiors, reducing manual review time.

• Spearheaded the design, development, and deployment of the full engineering pipeline for image feature extraction.

• Defined the data contract and schema for image label storage, ensuring consistency and downstream usability.

• Partnered with AI Platform to cache quality labels in Redis, enabling low-latency access during valuation workflows.

• Configured service-level AWS IAM roles for secure and reliable S3 read/write access.

• Facilitated sprint ceremonies and cross-team collaboration, increasing engagement and improving onboarding velocity.

Technologies: Python, AWS S3, Redis, IAM, Computer Vision

Deprecating Monolithic Library – Modernizing Human-Supported Valuation Models

Led the successful deprecation of a legacy monolithic library used in Zillow’s human-supported valuation models, improving system modularity, code quality, and long-term maintainability.

• Eliminated legacy dependencies, reducing technical debt and enabling faster, more reliable development across model teams.

• Improved test coverage and reliability by migrating critical functions into modular services and writing comprehensive unit tests.

• Streamlined codebase by consolidating functionality into shared libraries, reducing duplication and increasing reuse.

• Migrated essential code from the deprecated monolith into active projects, ensuring functionality parity and minimal disruption.

• Resolved key bugs caused by inconsistent column name suffixes, and implemented configurable overrides to enhance UI flexibility.

• Removed experimental and unused classes, reducing clutter and improving code maintainability.

• Fixed version conflicts by updating outdated dependency pins and aligning them across projects.