I build AI software that holds up in production.

Building an internal servicing assistant that classifies operations inquiries, retrieves approved policy context, and routes uncertain drafts through human review.

• ~30% less manual triage with Gemini classification
• ~92% release-gated classification accuracy
• Auditable retrieval, reviewer decisions, and confidence controls in PostgreSQL

Owned connected customer, employee, and admin applications with C#/.NET, Angular, TypeScript, SQL Server, and Azure.

• 2,000+ daily users across three applications
• 32% lower API latency after eliminating N+1 patterns
• 25% higher build reliability through root-cause analysis

Improved static-analysis detection, regression coverage, and data-intensive scan reporting for an application-security product.

• 25% higher file-detection accuracy with PHP and Bash support
• 25–35% faster MongoDB workflows over 10K+ JSON records
• Removed plaintext key storage through AES encryption and rebuilt key management