CV

Contact Information

Name Amir Hajian, Ph.D.
Professional Title AI Research & Product Exec
Location Toronto, Ontario

Professional Summary

Executive and scientist with 10+ years building and leading multi-disciplinary teams of researchers, engineers, and data specialists across academia and industry. Deep expertise in multimodal and generative AI, responsible AI (robustness, bias, safety, evaluation), and post-training adaptation of foundation models in enterprise and regulated contexts. Author of 70+ peer-reviewed publications (h-index 47); strong track record translating frontier research into scalable, trustworthy production systems—from wearables to agentic AI in financial services. Experienced in organizational planning, mentorship, and growing senior technical talent.

Experience

  • 2024 - present

    Toronto, ON

    Chief Science Officer & Head of Arteria Research Lab
    Arteria AI
    Built and currently heading up Arteria AI Cafe, a research lab for innovating novel techniques in multimodal agentic AI (NLP and computer vision) powered by GenAI. Leading science, engineering, MLOps and product teams to take cutting edge research to product.
    • Lead the design and building of DAX: a fully agentic suite of applications for all documentation tasks. 40x improvement to efficiency compared to the existing methods.
    • VLM/LLM post-training from research to production.
    • Full stack agentic AI app development: from infrastructure to launch in production.
    • Building powerful multimodal GenAI products based on in-house innovative research on agentic AI that learns in real time.
    • Leading research in cutting edge AI, with publications in major conferences including AAAI 2024 and The ACM WebConf 2025.
  • 2021 - 2024

    Toronto, ON

    VP of AI & Chief Scientist
    Arteria AI
    Built and managed the R&D organization consisting of teams of AI researchers, data scientists, ML Engineers, MLOps and annotators. Led to Arteria being named as the Most Promising 100 AI Startups by CB Insights and one of the 3 Coolest GenAI startups in FinTech by Gartner in 2023.
    • Led the end-to-end design and building of CUCINA: an enterprise-grade, compliance-ready multimodal document understanding system for regulated financial services.
    • Invented a novel multimodal architecture advancing the state of the art in vision-language modeling. Published in AAAI 2024.
    • Making use of Large Language Models (LLMs) and Generative AI for efficient information extraction.
    • Founded MLOps at Arteria for efficient productionization of large, complex deep learning models.
    • Mentored UWaterloo MS students on Multimodal and Computer Vision research.
  • 2019 - 2021

    Toronto, ON

    Director of Applied Research
    Scribd
    Founded Scribd’s Applied Research Team in Toronto to build products using machine learning to help customers explore the library, discover great content and read what they love.
    • Hands on AI product management: led the company’s efforts to create innovative AI-products using novel machine learning methods for search and recommendation.
    • Built end to end products, from ideation to production, leading distributed teams of researchers, engineers and project managers.
    • Created a superior user experience using the state of the art in computer vision, NLP, deep learning, statistical modeling and distributed computing in the cloud.
  • 2017 - 2019

    Toronto, ON

    Director of AI Research
    Thomson Reuters Labs
    Led company’s efforts to build AI-powered solutions at scale to create value and shorten time to market.
    • Designed and led the end-to-end development of an AI as a Service product in the cloud for creating a searchable video database using computer vision, deep learning, cloud computing and NLP.
    • Led company’s initiative in design and development of an end-to-end AI-powered Risk product from scratch. Patent pending.
    • Organized Machine Learning for Finance Professionals Workshop in collaboration with CFA Society, Toronto.
    • Organizer of Toronto Deep Learning Meetup since 2015.
  • 2015 - 2017

    Toronto & Waterloo, ON

    Senior Data Scientist
    Thomson Reuters Labs
    Founded the labs in Canada. SME and point of contact for data-driven product development. Built new solutions by rapid prototyping and iterative customer validation.
    • Initiated and built the scientific team in the labs.
    • Built 17 POCs in 18 months.
    • Successfully led projects from problem definition to production: Recommendation Engine, White House Run App.
  • 2010 - 2015

    Toronto, ON

    Senior Research Associate
    CITA, University of Toronto
    Designed Machine Learning algorithms to model the universe. Led large international collaborations to extract information from massive noisy data. Supervised research teams and students.
    • Spearheaded creating a modern scientific computing framework for all computing needs of the team.
    • Consulted a startup, solved data problems for building wearables devices.
  • 2006 - 2010

    Princeton, NJ

    Research Scientist
    Department of Physics, Princeton University
    Designed and built an end to end data analysis pipeline used as the backbone of a large collaborative effort to measure the properties of the universe based on multiwavelength observations using Bayesian statistics and probabilistic programming.

Education

  • 2006 - 2010

    Princeton, NJ

    Postdoctoral Researcher
    Princeton University
    Data Driven Cosmology
  • -

    Ph.D.
    Inter University Centre for Astronomy and Astrophysics
    Astrophysics
  • -

    Master of Science
    Institute for Advanced Studies in Basic Sciences
    Physics
  • -

    Bachelor of Science
    Sharif University of Technology
    Physics

Skills

Leadership (Master): Building and managing high performing, happy and diverse teams on the foundations of innovation, inclusivity, constant learning and collaborative team culture
Product (Master): Building powerful products using deep multimodal models combining computer vision and NLP
Computer Vision (Master): Vision Language Models (VLMs), Diffusion and Flow Matching GenAI, Object detection, text extraction, scene extraction and semantic segmentation in videos and images
NLP (Master): Transformers architecture, LLM based Language agents for solving complex problems
Responsible AI & Alignment (Master): Post-training and fine-tuning of foundation models, evaluation frameworks and benchmarks, robustness, bias/fairness analysis, safety, and explainability in enterprise and regulated contexts
MLOps (Master): Optimizing models and pipelines to use massive deep learning models in production at the lowest cost, least latency and shortest development time
Programming (Master): Python, PyTorch, JAX, Scala; cloud infrastructure on Databricks

Languages

English : Fluent

Interests

Multimodal Agentic AI: Multimodal AI, Agentic AI, GenAI, LLMs, VLMs, Post-training, AI for Science

Publications

Projects

  • DAX - Fully Agentic Documentation Suite

    A fully agentic suite of applications for all documentation tasks. 40x improvement to efficiency compared to existing methods.

    • Multimodal Agentic AI
    • NLP and Computer Vision
  • CUCINA - Multimodal Document Understanding

    Enterprise-grade, compliance-ready multimodal document understanding system for regulated financial services, extracting structured information from the most complex document layouts using Foundation Models.

    • Vision-Language Modeling
    • Foundation Models