Freddie Mac
Product Health Dashboard (PHD)
Division-wide platform - 8 Spring Boot microservices - giving engineering leads real-time visibility into application health, compliance, and cost metrics across all 10 SF Business Tech Ops branches.
- Built the Angular frontend from scratch and extended the Java / Spring Boot services behind it - new REST endpoints across the 8-service suite, delivering a dashboard adopted in production by leads across all 10 branches.
- Authored domain-specific scheduled batch jobs - each service ingesting from its own source system (Jira, Bitbucket, and others), then aggregating and scoring the results against company compliance and engineering standards before they land in the shared store.
- Diagnosed and remediated data-integrity bugs in the metric ingestion pipeline against PostgreSQL-backed services - scalable fixes that improved cross-service consistency and eliminated downstream scoring inaccuracies.
Stella - AI-Powered Internal Chatbot
Led end-to-end design and independent development of a Retrieval-Augmented Generation chatbot that guides users through the PHD and Starter Kit, answering natural-language questions and surfacing live health metrics on demand.
- Built a Python / FastAPI backend microservice and an Angular conversational UI; the implementation is now the internal reference architecture for future chatbot projects across both frontend and backend.
- Partnered cross-functionally with internal UX designers to align the chatbot with company design standards and accessibility requirements.
IDP Agent - intelligent document processing
A system that ingests mortgage lending documents from loan processors and uses AWS Textract with an AI layer to automatically parse borrower data points across a wide range of structured document types.
- Architected a Textract-based pipeline that extracts text, tables, and form fields from diverse structured documents.
- Built the AI normalization layer that parses borrower and loan data points across varying layouts - turning manual paper intake into clean, structured data.
Kiro Governance - AI IDE Steering & Hooks
Authored the steering documents and validation hooks for Freddie Mac's company-wide Kiro AI IDE rollout - so that as developers across every division adopt AI-assisted coding, the AI ships safe, on-standard, on-policy code by default.
- Authored steering documents that capture company coding standards, security policies, and architectural conventions - encoded directly into Kiro so guidance is enforceable inside the IDE, not buried in a wiki.
- Built validation hooks that run automated code-safety and linting checks against AI-generated code before it lands.
Starter Kit & Delivery Toolchain - company-wide platform tooling
The code generator and CI/CD platform every Freddie Mac team builds on - producing deployment-ready microservices and validating them before they ship.
- Re-engineered database connectivity in the generated service template - a ~300% increase in connection performance and request throughput for every team that consumes the kit.
- Implemented automated validation for duplicate gateway and ingress configurations, preventing misconfigured repositories from generating multiple gateways and improving deployment reliability across the org.