Karanjot Singh

AI engineering capstone - AISH AI-Assisted Application Portal - Engineered an AI document-extraction pipeline integrating Claude vision (via Google Vertex AI) with schema-enforced tool calling, converting ~20 document images per applicant into structured form data at ~85% field pre-fill. - Designed a concurrency-capped parallel processing flow that reconciles ~180 extracted field-value candidates down to a validated 73-field form, using a type-aware merge with per-field JSON-Schema validation. - Built a multi-model architecture pairing the vision-extraction model (Sonnet 4.6) with Claude Haiku for plain-language applicant guidance, plus an explainability layer attaching source evidence to every AI-populated field. - Implemented the system end-to-end across a Vue 3 frontend, Node/Express API, and PostgreSQL, with an automated Playwright test suite and security/accessibility checks.

Ministry of Technology & Innovation (Energy) Worked on AMI (Agreement Management Information System), a mission-critical legacy enterprise platform supporting Alberta’s Crown mineral agreement lifecycle. Contributed to Phase 1 enhancements and worked across C#, .NET Framework, WinForms, WCF, SQL Server, iBATIS.NET, AutoSys, TFVC/TFS, AzMan, and XML-based mappings to debug production and non-production incidents, reverse-engineer complex billing and workflow logic, and deliver controlled fixes in a highly integrated environment. Owned incident resolution across application, service, database, workflow, and batch-processing layers; supported builds across multiple environments; acted as a primary technical bridge between AMI Support, Production Operations, GIS, migration teams, business analysts, and stakeholders; and contributed to billing redesign work through effective-date rate logic, rollback-safe SQL migrations, and defect remediation.

CMPUT 379 - Operating System

● Developed new CRM components from scratch, ensuring the delivery of high-quality products. ● Optimized power automate flows to improve the scalability of production code. ● Conducted unit and smoke testing to uphold high standards and identify issues promptly. ● Mentored team members in CI/CD processes and promoted collaboration through consistent engagement.

● Implemented enhanced error capturing methods, improving error tracking and software quality. ● Resolved inefficiencies by optimizing API calls, improving workflow efficiency and reliability. ● Contributed to streamlining development processes and enhancing the user experience.