Saman Bazargani

AI Engineer @ Raven Connected

About

I build production-grade AI systems that operate at scale, from real-time edge perception to cloud-native ML infrastructure. Currently an AI Engineer at Raven Connected, I architect and deploy Edge AI platforms serving 100K+ devices, enabling real-time autonomous perception, V2X intelligence, and hybrid edge-cloud localization systems. My work spans multi-head computer vision models, INT8 optimization, DSP/NPU acceleration, and scalable AWS-based ML pipelines. I specialize in: • Edge AI & real-time inference • Cloud-native ML infrastructure (AWS, MLflow, SageMaker) • Distributed ETL & geospatial data systems • Hybrid localization (sensor fusion, pose graph optimization) • LLM-powered automation and agentic workflows I focus on building systems, not just models, ensuring reliability, scalability, and production performance across the full ML lifecycle. Open to opportunities in: Autonomous systems • AI infrastructure • Applied ML • Cloud-scale AI platforms • Graph Neural Networks • Knowledge Graphs • Reinforcement Learning • Multi-Agent Reinforcement Learning

Country

Canada

City

Ottawa

Industry

Computer Software

Skill

Robotics, Graph Neural Networks, Data Architects, Embedded Systems, Bash, CMake, Shell Scripting, Database Administration, Scrum, Operations, Resource Planning, Web Content, Business Applications, Design Documents, Solution Architecture, Diagram Design, Data Architecture, Vulnerability Management, Critical Thinking, Data Analytics

Experience

AI Engineer

Raven Connected

2024-8 - Present · 2 yrs 2 mos

Ottawa, Ontario, Canada

Developing machine learning models using computer vision techniques for (Embedded systems) microcontrollers.

Senior Research Associate

Carleton University

2024-2 - 2024-9 · 8 mos

Ottawa, Ontario, Canada

Studying mathematical modeling for online 3D rendering simulations of a LiDAR point cloud for firefighting training purposes using 3D Voxel Grids. Studying algorithms for efficient keyword search in large-scale graphs (100M+ nodes) by leveraging a hybrid KT-Index for streamlined retrieval.

Research Assistant

University of Ottawa

2019-9 - 2024-9 · 5 yrs 1 mo

Ottawa, Canada Area

Teacher

Shahid Dastgheib High School

2016-9 - 2018-9 · 2 yrs 1 mo

Designed and instructed a programming course emphasizing well-known algorithms, data structures, and graph theory problems to prepare talented students for national-level computer Olympiad competitions.

Head of ACM Association

Shiraz University

2016-9 - 2017-9 · 1 yr 1 mo

Education

University of Ottawa

Computer Science

2019-9 - 2023-10 · 4 yrs 2 mos

My Ph.D. thesis delves into the realms of computational geometry and graph theory. Within computational geometry, my research has revolved around the concept of transversals for geometric objects. Specifically, I've delved into a Helly-type problem where these objects are pairwise intersecting. Additionally, I've conducted a study on anagram-free coloring, where we've made some noteworthy findings. Specifically, our research revealed that the anagram-free chromatic number of graphs with a pathwidth of 2 is unbounded. We also introduced the concept of geodesic anagram-free coloring, and in this context, demonstrated that the anagram-free chromatic number of chordal graphs is bounded, particularly in relation to their pathwidth.