Jane Wang

Developed AI desktop companion with PyQt6 frontend and FastAPI backend, implementing RAG architecture with ChromaDB vector database and embedding-based semantic search. Implemented LLM-powered user profile extraction using conversation analysis to automatically extract personality traits, preferences, and goals for personalised AI interactions. Containerized microservices using Docker with multi-stage builds and orchestrated Kubernetes deployment with HPA auto-scaling and resource quotas. Configured infrastructure with ConfigMaps, Secrets, PersistentVolumeClaims, and health checks. Conducted load testing using Locust, simulating 50 concurrent users.

Built automated ingestion pipeline integrating PyPDF2, pdfplumber, and OCR to process multi-source PDFs; implemented text parsing to recognize question structure; designed unified JSON schema for multiple question types; created modular React components to enable batch Supabase uploads. Developed AI tutoring prototype using LangGraph and OpenAI API with multi-step conversation flow and personalized explanations; validated with 30+ beta testers before feature release. Built data pipeline using dbt, Airflow, and Snowflake to transform user answer data into analytics tables; optimized queries to reduce analysis time. Assisted in organizing in-person testing sessions to recruit users for question bank validation; collected feedback and fixed OCR errors through user incentives.

Designed and implemented RESTful APIs using Spring Boot and Spring Cloud to handle procurement and inventory operations, supporting over 15K SKU-level transactions per day; optimized inventory query performance via compound indexing and Java Streams; integrated Redis caching achieving 85% cache hit rate and reducing DB reads by 40%; implemented fuzzy search with Elasticsearch to improve product catalog lookups; set up CI/CD pipelines with Jenkins, Docker, and Kubernetes for deployment and scaling, delivering 99.9% uptime and reducing deployment time by 55%.