Thomas Assalian
BSc Computer Science — Concordia University, Montreal
Projects
2025
- Currently contains the compiler front-end only (a lexer + token generation system) for a custom language, producing standardized token streams for parsing.
- Includes validation rules + error reporting for malformed lexemes, reserved word misuse, and invalid identifiers.
PythonTheoretical Computer Science
- Built a Chrome extension that parses e-commerce product pages to extract fabric composition and compute a transparent quality score, helping users evaluate material value at the point of purchase.
- Implemented a weighted scoring engine, cross-site DOM scraping with content/background scripts, an AI chatbot, and an interactive UI with real-time sync and fabric insights.
JavaSciptOpenAIHTML/CSSWeb ScrapingBackend API
- Built a containerized agentic RAG service with agent orchestration, concurrent multi-tool execution, semantic caching, and MongoDB vector search.
- Improved end-to-end latency by 45% and retrieval relevance by 30% via caching + pipeline optimization.
PythonLangChainMongoDBDockerRedisGoogle GeminiCSS
2024
- Developed a React, AI agent web application that integrates Google Calendar with natural speech commands powered by Google Gemini and Web Speech API, reducing event management time by 62%
- Implemented a back-end pipeline to process user requests via Google Gemini API, outputting structured JSON to Google Calendar
ReactGoogle GeminiSupabase AuthWeb SpeechGoogle Calendar
- Developed a full-stack system enabling secure file uploads for 1,000+ employee profiles, reducing upload failures by 90% using S3-backed storage + metadata persistence.
- Implemented local + cloud storage support with AWS SDK, containerized deployment, and IAM access.
AWSJavaSpring BootReactDockerPostgreSQL