Alessandro Palmas ๐Ÿš€
Alessandro Palmas

Senior Research Engineer

About Me

Iโ€™m a Senior ML Research Engineer with 15+ years of experience building intelligent systems across foundation models, reinforcement learning, multimodal AI, and high-fidelity simulation. My work spans the spectrum from emerging research to production-scale engineering, with a focus on turning new ideas into scalable and reliable AI systems.

Currently, Iโ€™m part of the core research and engineering team at LawZero, a non-profit AI research lab in Montreal led by Yoshua Bengio that raised 300M+ USD to date, working on the next generation of foundation models and safe-by-design AI.

My current research focuses on large-scale post-training of foundation models with reinforcement learning, including distributed RL at hundreds-of-GPUs scale, verifiable training environments and rewards, reasoning models, model evaluation, and inference optimization. Iโ€™m particularly interested in the systems and algorithms required to make foundation models more capable, controllable, and reliable.

Before focusing on foundation models, I spent more than a decade developing intelligent systems across aerospace, defense, robotics, and gaming, with extensive experience in deep reinforcement learning, multi-agent systems, simulation, autonomy, and multimodal AI.

Iโ€™m particularly interested in the intersection of foundation models, reinforcement learning, and embodied intelligenceโ€”and in building systems that can reason, adapt, and ultimately understand the world rather than merely act in it.

Download CV
Interests
  • Foundation Models
  • Reinforcement Learning
  • Embodied / Physical AI
Experience
Education
  • MSc by Research (Post Graduate) Space Engineering

    University of Glasgow

  • MSc AeroSpace Engineering

    Turin Polytechnic

  • MSc Space Engineering

    Milan Polytechnic

  • Multidisciplinary International Program

    Alta Scuola Politecnica

  • BSc Aeronautical and Space Engineering

    Turin Polytechnic

Open Source Projects

Building is the fastest way to learn.

Experience Summary

  1. Member of Technical Staff

    LawZero - Montreal โ€ข Permanent Full-time
    Core member of the research and engineering team at LawZero, a non-profit AI research lab led by Yoshua Bengio, advancing safe, transparent, and interpretable AI systems that reason about the world.
    Skills: C++ ยท Python ยท PyTorch ยท vLLM ยท Docker ยท Foundation Models ยท Large Language Models (LLM) ยท Generative Flow Networks
  2. Senior Research Engineer

    Ubisoft La Forge - Montreal โ€ข Permanent Full-time
    Built and deployed advanced Deep RL systems for AAA games, including scalable multi-agent and curriculum-based algorithms. Led cross-site development of a distributed training pipeline and integrated AI models into proprietary engines for NPCs and simulations.
    Skills: C++ ยท Python ยท PyTorch ยท Deep Reinforcement Learning ยท Large Language Models (LLM) ยท Docker ยท Reinforcement Learning ยท Machine Learning
  3. AI/ML Consultant

    Artificial Twin โ€ข Freelance
    Led end-to-end AI/ML development across RL, LLMs, CV, and MLOps for real and simulated environments. Built and deployed transformer-based models, VLAM prototypes, and industrial CV systems, while maintaining core AI and geometry libraries.
    Skills: C++ ยท Python ยท PyTorch ยท Deep Reinforcement Learning ยท Large Language Models (LLM) ยท Computer Vision ยท Deep Learning ยท Computational Geometry ยท Vision Language Models (VLM) ยท Vision Language Action Models (VLAM) ยท Reinforcement Learning ยท Computational Fluid Dynamics
  4. Founder & Principal Research Engineer

    DIAMBRA (Acquired in Dec 2024) โ€ข Part-time
    Founded and led DIAMBRA, a global RL competition platform with leaderboards and Twitch integration. Built DIAMBRA Arena with Gym-compatible environments to support advanced RL research and seamless library integration.
    Skills: C++ ยท Python ยท PyTorch ยท Deep Reinforcement Learning ยท Docker
  5. Principal Research Engineer

    Nurjana Technologies โ€ข Permanent Full-time
    Designed and deployed AI for autonomous drones and space systems, including CV-based ISR, GNSS-denied navigation, and embedded deployment. Led development of ML algorithms for satellite tracking and multi-UAV autonomy, integrating sensor fusion and simulation-based training.
    Skills: C++ ยท Python ยท PyTorch ยท Computer Vision ยท Deep Learning ยท Deep Reinforcement Learning ยท OpenCV ยท Modeling and Simulation ยท Space Flight Dynamics
  6. AI/ML & Simulation Advisor and Industry Expert

    NATO โ€ข Contract Part-time
    Advised NATO on integrating AI/ML into defense systems for autonomous operations, sensor fusion, and decision-making under uncertainty. Contributed to AI-enhanced C4I for air defense and airdrop missions, focusing on real-time capabilities, trustworthiness, and emerging threat response.
    Skills: Artificial Intelligence ยท Machine Learning ยท Deep Learning ยท Computer Vision ยท Deep Reinforcement Learning ยท Flight Dynamics ยท Computational Fluid Dynamics
Papers
(2026). PyINE: A Framework for Scalable Elicitation and Oversight via Code Execution. arXiv preprint (cs.AI).
(2026). Language Models Recognize Dropout and Gaussian Noise Applied to Their Activations. arXiv preprint (cs.AI).
(2025). Application of Singular Perturbation Theory to Space Flight Dynamics Problems. Astrodynamics - Springer and Tsinghua University Press.
(2023). Performance Index of a Network of Ground-Based Optical Sensors for Space Objects Observation and Measurements. Advances in Space Research.