Saad Khalid
Thinking about frontier applications of AI agents: harnesses, reinforcement learning, and agentic system design. Broadly interested in human and machine cognition, engineering principles, design, culture and anthropology, and decision-making science.
Currently building ultracontext.co, a context-aware workspace that scales human thinking. Previously worked across application infrastructure, system design, and distributed systems. Early adopter of AI-augmented engineering. I write about my work here.
CS @ McMasterU
SWE @ Scotiabank
Previously
Technology Consulting @ EY
Infrastructure Engineering @ Shopify
Backend dev @ Aviato
Core dev @ PadawanDAO
North American finalist in Microsoft's Imagine Cup
Re-implemented the U-Net paper
Shipped an anonymized chat app at Buildspace Nights and Weekends
Developed CNNs from first-principles. Work can be found here.
Current Projects
ultracontext.co
A workspace to scale human thinking — agents handle the busywork, you just make decisions
Agents
Documenting my self-directed learning journey into AI agents
Simple Autodiff
Autodiff implementation from first-principles
npager.com
Generate and share variable length briefing docs from a single prompt
envm
CLI tool for managing .env files — diff, merge, snapshot, validate
Writing
Links
More about me →