Multi-Agent Systems
LangGraph and LangChain orchestration across specialized agents for call analysis, proposal generation, and task routing — over GPT, Claude, Gemini, and LLaMA.
Computer Science @ York University · AI Engineer
I build AI agents that actually ship.
I build multi-agent systems, custom MCP servers, and RAG pipelines — and put them in front of real users, not just in notebooks.
What I work with
Systems, retrieval, and what runs them.
RAG & Retrieval
Tree-sitter AST chunking into ChromaDB for cited, hallucination-resistant Q&A over codebases of up to 10,000 files.
Backend & Deployment
A 5-service backend on Railway with async SQLAlchemy and Celery, atomic Redis rate limiting, and 230+ automated tests behind it.
Selected work
Things I built & shipped.
SixRise
PythonFastAPITypeScriptShopify Admin APIpgvectorRedis
A live Shopify embedded app that lifts merchant visibility inside AI shopping assistants — auditing catalog data, publishing grounded fixes, and measuring share-of-model against named competitors.
- Eight-node async agent pipeline fans buyer-intent queries across Perplexity Sonar and OpenAI, parses cited brands, and computes per-engine share-of-model.
- Grounded Optimizer with a negative-grounding guard and two-layer staleness gate — never fabricates attributes, and routes every Shopify Admin API write through preview/approval.
- TypeScript/Python monorepo on Northflank, Neon, and Upstash Redis with split-schema ownership; fixed a connection-pool advisory-lock deadlock via transaction-bound session storage.
ContextCode
FastAPICeleryChromaDBPostgreSQLRedisNext.js
A full-stack RAG system that indexes GitHub repositories for cited, hallucination-resistant Q&A across codebases of up to 10,000 files.
- Tree-sitter AST parsing for function- and class-level chunking across Python, JavaScript, and TypeScript.
- Cloudflare Turnstile plus atomic Redis dual rate limiting, per-session and global — a public zero-signup demo with bounded API spend and bot protection.
- 5-service backend on Railway with async SQLAlchemy and SSE progress streaming, a Next.js frontend on Vercel, and 230+ automated tests.
AI Stock Insights
StreamlitTechnical IndicatorsEnsemble Models
ML-powered stock analysis with technical indicators and ensemble prediction models, served through a Streamlit dashboard.
RateFlix
Java SwingMySQLTMDB
A movie-review platform with authentication, watchlists, and TMDB integration, built in Java Swing over MySQL.
Experience
Where I’ve worked so far.
- Built multi-agent AI systems using LangGraph and LangChain for automated call analysis, proposal generation, and task orchestration across specialized agents, integrating GPT, Claude, Gemini, and LLaMA models.
- Developed 10+ custom MCP servers connecting AI agents to external APIs, databases, and business applications, including tool design, authentication, and deployment.
- Engineered Retrieval-Augmented Generation (RAG) pipelines for context-aware AI assistants and document-intelligence workflows, enabling knowledge retrieval over business data.
- Created event-driven lead-generation automation with n8n, integrating LLMs, CRM platforms, and web scrapers to streamline prospect identification and reduce manual outreach effort.
About
A little more about me.
I’m a Computer Science student at York University, finishing in Dec 2026. Most of what I build sits where LLMs meet real infrastructure — agent pipelines that hold up in production, retrieval systems that cite their sources, and the unglamorous backend work that keeps both honest.
Education
York University — BSc. Honours in Computer Science
Toronto, ON · Expected Dec 2026
Coursework
Machine Learning, Artificial Intelligence, Data Structures & Algorithms, Databases, Operating Systems, Computer Networks, Software Design.
CS Hub — Computing Students’ Hub
Events & Academics Team, Toronto, ON