Louis-François Bouchard, known as What’s AI, is cofounder and CTO of Towards AI. His teaching connects language-model research with the evaluation, retrieval and cost decisions needed to turn an application into a useful production system.[1][4]
Start with Context Engineering in 2026: Compaction, Memory & Cost →
Education
Doctoral studies in AI and computer vision; left without completing Ph.D.
Completed 2024
[1][4]
Links
Owned professional biography
Owned linked GitHub account Original conference career biography 
X · whats_ai Cofounder and Chief Technology Officer · Towards AI
Creator · What’s AI
Career
Bouchard studied and worked in AI in Montreal before concentrating on Towards AI. He pursued doctoral research at Mila and Polytechnique Montréal and left that path to build and teach. Earlier, he led AI development at designstripe. His What’s AI channel began with research explanations; his later work spans courses, practical guides and books on building language-model applications.[1][4]
Contributions
Choose memory policies against the actual constraint
Their workshop compares full history, compaction and retrieved memory against recall, latency and cost. Shortening prompts can discard facts or break cached prefixes. Prompt caching affects serving cost, while the context window limits input capacity.[5]
Build a learning path around complete applications
His AI-engineering guide organizes resources around language models, prompting, retrieval, agents, evaluation and deployment. It combines linked lessons with a suggested progression so a learner can connect isolated techniques to an application. Bouchard’s broader teaching favors a clear workflow and measurable evaluation before adding autonomous or multi-agent behavior.[6][1]
Tools, writing & talks
Writing, research & talks
Start here
A workshop comparing context policies through recall, serving cost and caching.
Compare full history and compaction against recall and caching. Familiarity with context windows and retrieval helps separate capacity from cost.
Co-presenter with Omar Solano and Samridhi Vaid[5]- Context Engineering in 2026 — Louis-François Bouchard, Omar Solano & Samridhi Vaid, Towards AIyoutube.com
Across 11 context presets measured on an open source AI tutor, sending the full history beat every compaction technique on recall, cost and latency at once, because rewriting context invalidates the prompt cache that serves most of the tokens.
Catalog workOriginal source
Tools
- Start AI Engineeringgithub.com
A curated progression of practical resources for building, evaluating and deploying AI applications.
Creator and curator[6]