Yantian Yin

Education

McGill University

B.Sc. Honours Computer Science

GPA 3.7+ / 4.0

Montreal, Canada

2023 — April 2027

Tencent

Tencent Cloud Business & Product Trainee (Key Account)

Analyzed AI companies with high compute demand and built targeted outreach strategies
Proposed cost-optimization solutions (GPU pricing, SaaS/PaaS adoption) for startups and enterprises
Organized networking sessions and facilitated meetings with enterprise and government stakeholders
Led a government PoC end-to-end: defined requirements, built demo environments, and presented solutions

Shenzhen, China

Jun – Sep 2025

Campango

Software Developer

Identified performance bottlenecks caused by repeated Firestore queries for post retrieval
Designed and implemented an in-memory caching layer with TTL to reduce redundant database access
Conducted testing and debugging to ensure correctness and reliability of backend functionalities

Campus — Student Social & Academic Networking Platform

React Native, Node.js, PostgreSQL/Supabase, AWS

Developed a mobile social platform to help students connect with classmates and share academic and career-related content
Built a recommendation system by parsing unofficial transcripts to match users with relevant classmates and communities
Implemented real-time messaging and authentication using a Node.js backend and Supabase services
Designed and optimized PostgreSQL schema for scalable data storage and efficient querying
Applied a content classification model to detect and prevent inappropriate posts

Value — iOS Second-hand Marketplace & Depreciation Calculator

Swift, Firebase · Supervised under Joseph Vybihal · github

Second-hand marketplace app with functionality to estimate daily cost of ownership and depreciation curves for electronics
Implemented real-time in-app messaging between buyers and sellers using Firebase
Optimized Firestore data model with composite indexes, batched writes, and denormalization
Designed an exponential-decay model to estimate price depreciation and highlight early-adopter premium

PyTorch Checkpoint Tracing & MLCommons Benchmarking

Research · Supervised under Oana Balmau

Prototyped PyTorch hooks to capture checkpoint save/load metadata
Integrated checkpoint events into execution-trace schema (Chakra benchmarking module)
Implemented ET-Replay/Chakra-side simulation for checkpoint operations via I/O emulation
Enabled neutral, reproducible storage benchmarking on real ML training traces