Essam Sleiman

Essam Sleiman

I run Canvas Labs (YC), where we work on self-improving AI. We recently open-sourced meta-agent, a framework for agent continual learning.

Previously, I did research on continual learning, language-model interpretability, and vision-language models at Amazon Science and as a grad student at Harvard, with papers at ECCV and CVPR. In my free time, I enjoy basketball, coffee, and hiking around SF.

Email GitHub Scholar X LinkedIn

Selected Work

67%
→
80%
→
87%
Founder, Canvas Labs (YC F24) Meta-Agent: Open-Source Framework for Agent Continual Learning
1. Continual Learning for Agents Mar 14 blog 2. Reward Modeling as Harness Optimization Apr 22 blog 3. AutoHarnessBench coming soon
blog code agents self-improving ai
Research Engineer at Amazon Science Continual Self-Supervised Learning with Knowledge Distillation Essam Sleiman*, Xiangbo Li, Saad Ali · Preprint, 2023 Continual self-supervised learning for Twitch video embeddings, leveraging knowledge distillation to prevent catastrophic forgetting across time periods.
paper continual learning self-supervised
Goldfish long-video retrieval framework architecture
Research, Harvard Goldfish: Vision-Language Understanding of Arbitrarily Long Videos Essam Sleiman* (Contributor) · ECCV 2024 Vision-language understanding for arbitrarily long videos using retrieval to select relevant clips before answering questions.
paper code vision-language long video
Research, Harvard & MIT Do Function Vectors Factor Task and Distribution? Essam Sleiman*, Jacob Andreas · Preprint, 2024 Studying whether in-context learning representations in LLMs can be decomposed into separable task and distribution components.
paper code interpretability in-context learning
SlowFormer energy attack on vision transformers
Research, UC Davis SlowFormer: Adversarial Robustness for Efficient Vision Transformers Essam Sleiman*, K L Navaneet*, et al. · CVPR 2024 Adversarial robustness for efficient vision transformers: small universal patches break adaptive inference, while adversarial training helps recover robustness.
paper code robustness vision transformers

→ See all projects