Kien Nguyen

Improved grocery product search relevance for an AI shopping assistant used across 5+ Canadian retail chains, reducing irrelevant search results to near-zero as validated by 30+ automated regression tests, by building a hybrid BM25 + semantic vector reranking engine with intent classification and multi-signal scoring Designed a 4-stage NLP pipeline (normalization → retrieval → ranking → filtering) combining BM25 lexical scoring, FAISS ANN dense retrieval with 384-dim sentence embeddings over 258K products, and Reciprocal Rank Fusion to balance keyword precision with semantic recall Developed a rule-based intent classifier (9 categories) and grocery product taxonomy that enabled the ranker to discriminate raw ingredients from derived/processed products

Machine Learning Intern - SAVR – Smart Grocery Shopping - Mississauga, ON

(2026-02 - 2026-05)

• Improved grocery product search relevance for an AI shopping assistant used across 5+ Canadian retail chains, reducing irrelevant search results to near-zero as validated by 30+ automated regression tests, by building a hybrid BM25 + semantic vector reranking engine with intent classification and multi-signal scoring
• Designed a 4-stage NLP pipeline (normalization → retrieval → ranking → filtering) combining BM25 lexical scoring, FAISS ANN dense retrieval with 384-dim sentence embeddings over 258K products, and Reciprocal Rank Fusion to balance keyword precision with semantic recall
• Developed a rule-based intent classifier (9 categories) and grocery product taxonomy that enabled the ranker to discriminate raw ingredients from derived/processed products

Poll Clerk - U.T.M. Student Union - Mississauga, ON

(2024-11 - 2025-03)

• Automated ballot generation for 3,000+ voter election using Python (ReportLab, PyPDF2), reducing production time from 5 days to under 10 seconds.
• Led ballot-making operations by training staff while partnering with the C.R.O. to audit voting data through Excel integrity checks for anomaly detection.

Data Analyst Intern - Sky Fitness Yoga - Hanoi, VN

(2023-05 - 2023-09)

• Built Python ETL pipeline automating commission-based payroll and sales dashboards (1K+ monthly records, 30+ employees, 2 locations), reducing processing time from 3 weeks to 15 minutes and eliminating 24K annually.
• Developed a custom analytical suite using Python and Pandas to visualize monthly sales metrics, cutting weekly data analysis time by 80% and surfacing key revenue drivers

Honors Bachelor of Science - Applied Statistics, Computer Science, and Mathematics - University of Toronto (2026-06)