Kunal Saluja
Computer vision engineer with a decade of experience shipping production AI across AR/VR, robotics, retail, agriculture, and Security/Surveillance. I lead computer vision and AI teams that turn research into systems that scale.
Latest
Introduction to Reinforcement Learning
Slides walking through RL foundations and algorithm families, with Q-learning and PPO worked end to end on one grid world.
Towards LLM-Centric Multimodal Fusion
Slides on a survey of 125 multimodal LLMs, read through one question: how does a non-text signal get into the language embedding space?
LLaVA — Visual Instruction Tuning
Slides on LLaVA, the paper that wired an LLM together with a vision encoder using GPT-generated instruction-following data.
Illusion of Thinking
Slides on the "Illusion of Thinking" paper and what it implies about how today's reasoning models actually reason.
Selected work
All projects →
Security & Surveillance
License-plate recognition and natural-language video search for law enforcement.
3D Spatial Awareness
Vehicle dent detection via fused RGB + depth.
Pipe Inspection
Sewer-pipe anomaly segmentation across 100M ft of footage.
Retail Monitoring
In-store object and person detection at scale.