SINCE 2025
Aster AI
Founder & CEO
- On-Device Real-Time 3D Perception
- Digital Twins of Physical Space
- Full-Stack CV, Silicon to Model
- IC-to-Head-of-AI Applied Research
Computer Vision Engineer · Founder, Aster AI · Austin
Giri Murali
Austin, Texas
He turns messy physical spaces into live digital twins that run on the device itself.
Giri Murali founded Aster AI in 2025 to build the agent layer for professional life. Before that he spent thirteen years in physical-world computer vision: seven at Aquifi shipping real-time SLAM and depth in production C++, then a climb from engineer to Director of applied research at HOVER, then Head of AI at RADAR. He works the whole stack — hand-tuned SIMD kernels up through graph neural networks and differentiable rendering.
- 13+ yrsOn-device computer-vision depth
- 7 yrsLead CV/ML engineer at Aquifi
- 3 rolesHOVER: Senior Engineer to Director
HAPPENING NOW
STANDING TODAY
SINCE 2025
Syelo Ventures
CTO / Technical Advisor
HOW THEY THINK4 BELIEFS
Four beliefs carry across thirteen years and five domains.
THE IMPACT
Thirteen years of work, most of it shipped deep inside products.
EVERYTHING ELSE · 1 WORK
THE TIMELINE
One instinct across five domains — build a live, structured model of a messy real-world system.
The whole record, newest first – tap any moment and Aster tells the story.
2025-PRESENT
The domain jump — modeling work itself
In February 2025 Giri founded Aster AI as cofounder and CEO to build the agent layer for professional life — his first company after thirteen years in physical-world computer vision. Since June 2025 he also advises Syelo Ventures on AI strategy. The object of modeling shifts from houses and stores to professional work, carrying the same systems discipline forward.
“Building the agent layer for professional life.”
2023-2025
Head of AI — a live twin of the store
As Head of AI at RADAR (April 2023 – January 2025), Giri owned core tech and applied research across computer vision, machine learning and data science to automate inventory, analytics and checkout for physical retail. The product: a real-time digital twin of dynamic stores, modeling geometry, people and products together, running on custom multi-camera hardware ported onto NPUs.
2019-2023
IC to Director — 3D twins of homes
Across three roles at HOVER (June 2019 – March 2023), Giri worked on structured semantic 3D reconstruction of residential properties from a sparse set of smartphone photos, with real-time on-device capture. The pipeline fused wide-baseline matching via deep reinforcement learning with graph neural networks and differentiable rendering. Two promotions in under four years carried him from IC to Director of the Applied Research Group.
“Structured semantic 3D reconstruction of residential properties with an intuitive consumer-grade capture experience leveraging real-time on-device scene understanding.”
2012-2019
Seven years at the edge — SLAM, VIO, depth on device
For seven years (July 2012 – June 2019) Giri was Lead Computer Vision & Machine Learning Engineer at Aquifi, his longest tenure and the bedrock of his edge-CV reputation. He built RGBD SLAM, ORB-SLAM, VIO, stereo and trinocular depth, and real-time deep learning — shipped in mission-critical production C++, hand-optimized with SIMD and GPU/OpenCL for Intel and ARM.
2011-2012
Academic origins — 3D motion tracking
As a Graduate Research Assistant in Computer Vision at the University of Michigan (June 2011 – March 2012), Giri worked on articulated 3D human-motion tracking in monocular video via parts-based template matching. In parallel he took his first industry role at Smiley Media on face detection and classification. The seed of a career spent recovering 3D structure from impoverished 2D input.