Machine Learning Engineer | AI Researcher | Deep Learning Specialist

Hello! I’m Arman, a Machine Learning Engineer with 5+ years of experience specializing in Large Language Models, transformer architectures, computer vision systems, and autonomous robotics. I focus on building production-ready AI solutions that bridge research and industry applications.

Professional Focus

My expertise spans across enterprise AI solutions, where I’ve implemented production-ready RAG systems reducing operational costs by 35% through PEFT/LoRA fine-tuning and quantization techniques. I’ve built retrieval systems with 92% accuracy using vector embeddings and hybrid retrieval approaches, developed multimodal pipelines integrating vision models with transformers, and deployed end-to-end ML infrastructure across major cloud platforms.

Research & Development

My research interests center on robot learning and autonomous navigation, with a core focus on developing self-adaptive neural networks for robust autonomous systems:

Core Research Areas:

• Visual-Inertial SLAM: Multi-modal perception and real-time state estimation for navigation under uncertainty
• End-to-end Learning: Real-time navigation in dynamic and uncertain environments
• Robust Sensor Fusion: Adaptive decision-making in unstructured terrains
• Vision-Language Models: Semantic scene understanding for autonomous systems
• Large Language Models: Interpretable navigation and mission planning

Current Research Direction: Developing self-adaptive neural networks combining:

• Vision-language models for semantic scene understanding
• Learning-based multi-sensor fusion for unstructured environments
• Learning-based SLAM optimization with physics-informed approaches

Recent Work: Transformer-Enhanced Physics-Informed Neural Networks (TE-PINN) achieving 36.8% error reduction in attitude estimation, submitted to ICRA 2025. This work integrates multi-head attention mechanisms with physics-based constraints (quaternion kinematics, rigid body dynamics) for robust real-time orientation estimation in high-noise, dynamic conditions.

Core Competencies

LLMs & NLP: PEFT/LoRA fine-tuning, 4-bit/8-bit quantization (GGUF, AWQ), RAG systems, prompt engineering Computer Vision: Object detection, segmentation, vision-language modeling (YOLO, Mask R-CNN, Vision Transformers) Autonomous Systems: ROS2, ORB-SLAM3, LiDAR-SLAM, visual-inertial odometry, multi-sensor fusion Data Infrastructure: Vector databases (FAISS, Qdrant, Vespa), BM25/TF-IDF indexing, hybrid retrieval systems ML Frameworks: PyTorch, TensorFlow, HuggingFace, Docker, FastAPI, TorchServe, DeepSpeed

Explore my professional experience and research projects in detail by visiting Experiences.

Research Collaborations: I am actively seeking opportunities for academic research collaborations in robot learning, vision-based navigation, and LLM applications for autonomous systems. 🤝💡

Selected Achievements

• Published Researcher: Papers in Elsevier Measurement (IF 5.2), submitted to ICRA 2025, BMC Medical Education
• Competition Winner: Ranked 1st in Oxford ML Summer School Health & Medicine track (120+ participants)
• MIT IAIFI Program: Research participant in AI+Physics program focusing on simulation-based inference
• Production AI: Deployed RAG systems processing 5M+ documents daily, achieving 92% retrieval accuracy
• Autonomous Robotics: Implemented VIO-LiDAR fusion achieving 15cm localization accuracy in warehouse environments
• Academic Service: Reviewed 100+ papers for IEEE, Elsevier journals; Outstanding Reviewer Award (2023)

Teaching & Mentoring

• Co-Instructor @ Tehran University of Medical Sciences: Application of Technology in Research
• Teaching Assistant @ University of Tehran: Fuzzy Logic (Graduate Level)
• Thesis Supervisor: 5 undergraduate theses, 1 master’s thesis in AI and robotics
• Mentor @ Space Generation Advisory Council: Guiding professionals in computer vision and AI

Feel free to reach out for research collaborations or to discuss AI/ML projects.