Elorian is hiring to solve open visual thinking problems in AI
AndrewDai · x · 2026-08-26
Elorian is hiring to tackle the open problem of teaching AI to think visually, enabling models to understand and reason through the visual medium natively like humans.
Core Arguments
- Visual Precedes Language: Humans understand the world through sight and touch before speaking; visual thinking is foundational for design, science, and coaching.
- Current Limits: Existing Vision Language Models (VLMs) rely on a fragile two-step process (translating visuals to text then reasoning), which is prone to hallucination and struggles with indescribable visual tasks.
- Generation vs. Reasoning: The ability to generate photos/videos does not translate to reasoning about visual content; there is a widening gap between coding and visual task performance.
Vision: To build models that understand spatial relationships, physical constraints, design intent, and abstractions to achieve complex solutions in the physical world.
More from Companies & People
- Startup Insight: Engaged User Feedback Thread is Most Valuable Asset — josh_bickett · 2026-08-26
- Prof. Mohit Banerjee to discuss Trustworthy Collaboration & Long-Horizon Memory at UCF AI Institute — mohitban47 · 2026-08-26
- Handot Expands Partnership with Microsoft to Serve More Ecosystem Products — handotdev · 2026-08-26
- Dev: Build for 6 months from now, ignore minor model diffs — zeeg · 2026-08-26
- Shopify Design Case Study: 60% of Designers Now Push Code to Production — floguo · 2026-08-26
- Nvidia Partners with Perplexity AI to Run on DGX Cloud — dd32x · 2026-08-26