I'm a computer science graduate with about 10 years of experience in Japan's automotive industry. My work has mainly involved embedded software for HMI systems, battery control software, CAN, and related vehicle communication technologies. I'm considering a move into AI and may eventually look for opportunities in Europe or Australia. Since my current background seems relevant to Edge AI, I'm wondering whether I should focus first on AI theory and fundamentals or spend more time building practical applications, agents, and deployable projects with modern tools. I'm not aiming to become an ML researcher, but I also don't want to rely on APIs without understanding the basics. I'm also concerned about how my Japanese and English proficiency could affect my options, as well as work visas and long-term career prospects outside Japan. I'm currently around 35, have a highly skilled visa in Japan, and don't have dependents, so I have some flexibility to relocate.
3 Answers
Your automotive and embedded background could be a strong route into Edge AI rather than starting over as a general AI applicant. Experience with vehicle communication, hardware constraints, HMI, and battery systems is difficult to find in new graduates. I’d focus on deploying models on constrained devices, C/C++ and Python, Linux, model quantization, inference optimization, computer vision or sensor data, and hardware accelerators. A few projects that connect AI to automotive or industrial use cases would probably help more than studying theory in isolation. Japan may have good demand for this combination, although relocating abroad will also depend heavily on employer sponsorship and visa rules.
The advice will depend on details such as your education, specific responsibilities, citizenship, visa status, and what you mean by working in AI. Based on the information you provided, your computer science degree and decade of embedded automotive experience are valuable assets. Rather than presenting yourself as someone starting AI from zero, position yourself as an experienced embedded engineer moving toward intelligent systems. Check immigration requirements early, because finding a job abroad and obtaining permission to work are separate challenges. Your language skills may limit some local roles, but international engineering teams often work primarily in English, so improving technical English and practicing interviews could make a significant difference.
That’s fair. I’m currently working on embedded HMI software and previously worked on battery control software, with most of my experience involving CAN and related communication systems. I’m in Japan on a highly skilled visa and have no children, so I’m relatively flexible. I’m mainly exploring alternatives because I’m uncertain about my long-term prospects here.
It helps to separate the paths you’re comparing. Developing new models is heavily based on mathematics, statistics, optimization, and research, while building products with existing models is closer to software engineering: APIs, backend systems, data pipelines, evaluation, and user-facing applications. Neither path is automatically better paid; compensation depends more on your skill level, business impact, and location. Given your experience, applied AI or Edge AI engineering seems more realistic than competing directly for research roles. Learn enough fundamentals to understand training, inference, overfitting, embeddings, evaluation, and common failure modes, then spend most of your time building and deploying useful systems.
I’m leaning toward the practical side and don’t plan to become an ML researcher. I’d still like enough theory to understand what’s happening underneath instead of just calling an API. My main concern is choosing a direction with solid long-term opportunities in places such as Europe or Australia.
That makes sense. Edge AI does feel like a more natural transition because of my embedded software and vehicle communication experience. For getting started, would you prioritize model deployment and optimization first, or should I build a stronger foundation in machine learning theory before working on projects?