Software Engineer | Technical Sales and SDR

I pitched CustomerMaps, won the company as a client, and built an agentic data-analysis system that enables non-technical users to query CSV data in plain English. The platform used a stateless TypeScript orchestrator to route requests to deterministic Python tools through JSON-RPC and MCP.

I reduced hallucination-induced pipeline failures by 85% by constraining the LLM to select one of five strategies while programmatically locking downstream agent sequences. I also reduced LLM API costs by 40% through a zero-LLM upload cleaning pipeline for deduplication, imputation, outlier handling, and PII processing, with hard termination guardrails, execution audit logs, hierarchical span telemetry, and pre/post-condition contracts.

I led the client build of the CareAble mobile caregiver marketplace MVP. I ran iterative prototyping and stakeholder feedback sessions, translating client needs into testable user stories, data models, and API contracts.

I developed an AI chatbot that classifies user intent, retrieves care preferences and recurring routines, and routes users to scheduling actions through MCP. I also implemented a PostgreSQL booking state machine with atomic transitions, time-based expiry, and row-level security to prevent race conditions, slot hoarding, and unauthorized data access.

I served as a front-line point of contact for more than 1,000 student residents in UBC Student Housing.

I applied customer service, communication, and issue-resolution skills while supporting a diverse residential community.

I worked in a front-of-house customer service role in Vancouver, British Columbia.

I developed face-to-face customer interaction, communication, and service delivery experience.

I built quality-of-life tracking and self-management features for PolarUs, a CIHR-funded Android application that helps users with bipolar disorder monitor symptoms more consistently and has more than 1,000 downloads.

I reduced redundant network requests by 70% and improved loading for more than 200 resources by creating a serverless Node.js OpenGraph API with Firebase Cloud Functions and a Firestore cache. I also fixed layout clipping across more than 10 Android device configurations using Window Insets in Jetpack Compose.

I developed capabilities for a RAG-based question-answering system, enabling PDFs exceeding 50 pages to be summarized within LLM context limits through recursive character text splitting with LangChain.

I improved summary quality by ranking document segments with TF-IDF and validating results against a curated benchmark of more than 25 documents. I also identified eight defects before deployment and eliminated manual regression testing by creating a pytest suite with more than 40 cases for API endpoints and data preprocessing.

Graduated with a Bachelor of Applied Science in Computer Engineering. Received the Outstanding International Student Award.