Introduction
Hi there! My name is Igor Sadalski and I'm originally from Poland. My passion for robotics and artificial intelligence started at a young age. I was that kid who was always tinkering with electronics and dreaming about building intelligent machines. So, pursuing a career at the intersection of engineering and computer science felt like a natural fit for me. To chase my dreams, I moved to the UK to study Mechanical Engineering with a focus on Mechatronics at Imperial College London. It was an incredible experience that exposed me to so many fascinating topics - from embedded systems and control theory to machine learning and reinforcement learning. I threw myself into projects, like leading the first ever team at my university developing fully autonomous racing car! One of the highlights of my undergraduate years was receiving a scholarship and spending two summers as a research intern at CalTech in Pasadena, US. Suddenly I was working alongside some of the top minds in robotics at one of the most prestigious research institutions in the world. It was surreal! I got to work on some really cool projects, like optimizing the control algorithms for a hopping robot and developing an AI system to enable safe drone flight in harsh conditions. Those experiences cemented my resolve to pursue a career in research and push the boundaries of what's possible with AI and robotics. After my undergraduate studies, I specialized further doing a master's in computer science at Imperial College London, specializing in AI and Machine Learning. Recently I just finished my placement at Harvard University working on my master's thesis! There I collaborated with Prof. Stephanie Gil to develop AI-powered systems that could pro-actively route shuttle buses at Harvard. The goal of my work was to leverage cutting-edge techniques like transformers to predict demand and enable smarter, more efficient allocation of resources like buses. Currently I work at Somite.ai, a AI-biotech startup, where we try to build foundational model for biology and cure Duchenne muscular dystrophy! It is amazing and very meaningful work. Specifically, I work as an AI Data Scientist/Research Engineer. My work consists mainly of developing new AI models for improving stem cell therapies. So, that's a little bit about me and my path so far. On a day-to-day basis, you'll usually find me in front of my computer, writing code in Python and running experiments with deep learning models, mainly transformers, and collaborating with my lab mates. There's nothing quite like the thrill of finally getting a complex AI system working after weeks of debugging and prototyping! Beyond research, I love cooking and spending time with my friends and family! Looking ahead, my ultimate goal is to use AI that can help solve real-world challenges be it biology, engineering or research. Whether it's large biological models, intelligent transportation systems, new smart materials design, robotic control design, I believe AI has immense potential to improve people's lives when developed thoughtfully and responsibly. Lastly, if there are any students out there who share my passions, I'd encourage you to start getting hands-on experience as early as possible. Seek out research opportunities, learn to code, join robotics clubs and competitions. And don't be afraid to reach out to people in the field - we're usually more than happy to chat and share advice! AI is such an exciting, fast-moving fields and there are so many amazing opportunities out there!
Mechanical Engineering Fellow
career options
Mechanical Engineering gives you a very broad exposure to different engineering fields and thus you can find yourself pursuing varied career paths!
Mechanical Engineering Fellow
skills
What are the main hard skills you use on a daily basis in your current job?
Python is my go-to programming language for almost all of my work. Its simplicity, versatility, and extensive ecosystem of libraries (i.e. small packages you can download and use to solve some specific problems) like numpy, pandas, scipy, matplotlib make it a powerhouse. I use Python daily for tasks ranging from data preprocessing and analysis, to implementing and training machine learning models. I first learned Python in my introductory programming courses and have deepened my skills through countless projects and reading multiple technical books.
PyTorch is my preferred python library which enables me to design deep learning models. I love its native Python structure, and detailed low-level control. I use PyTorch daily to develop, train, and deploy deep neural networks for training the biological model. I first picked up PyTorch skills through deep learning coursework and research projects and have since used it extensively in both academic and industry settings. PyTorch's flexibility and performance make it an essential part of my day-to-day AI engineering workflow.
As an open-source operating system, Linux is the foundation for much of the software engineering and infrastructure. I use Linux (usually Ubuntu) as my daily driver OS, and I'm very comfortable with the Linux command line for tasks like filesystem navigation, process management, remote access, and bash scripting. I first encountered Linux in university computer labs and on research computer clusters and have been using it daily ever since. Familiarity with Linux is, in my experience, a must-have skill for any serious practitioner in AI.
What are the main soft skills you use on a daily basis in your current job?
As an AI researcher and engineer, I constantly communicate complex technical concepts to diverse audiences, from fellow researchers to business stakeholders to the general public. This requires adapting my language and level of detail to ensure understanding without sacrificing accuracy. I honed these skills through presenting at conferences, writing technical blogs and papers, and collaborating on cross-functional teams. On a daily basis, I employ communication skills in situations like explaining my work to colleagues, documenting my code and results, and interfacing with other departments.
AI is a highly interdisciplinary field and almost all projects require close collaboration with experts from different domains. I've learned the importance of active listening, empathy, and finding common ground to foster effective teamwork. I gained these skills through experiences like leading the autonomous racing car team, which required coordinating mechanical, electrical, and software engineers. I use collaboration skills daily in meetings, pair programming sessions, and when providing or receiving feedback on work.
The pace of progress in AI and robotics is breathtaking, with new techniques and tools emerging constantly. To stay at the forefront, I've cultivated a mindset of continuous learning - always seeking out new knowledge and skills. I do this through reading papers, attending conferences and workshops, taking online courses, and engaging in inside projects. I apply this learning daily by experimenting with new approaches, updating my codebases with the latest best practices, and sharing insights with colleagues.
Igor
’s personal path
Tell us about your personal journey in
Mechanical Engineering Fellow
:
Looking back on my own journey from education to employment, one thing stands out: landing your dream job is a process, not an event. It takes time, perseverance, and often a good measure of luck. When I started my job search, I thought my impressive research experience and publications would make me an instant catch. I soon realized that while those credentials opened doors, they didn't guarantee a smooth path. I submitted countless applications, often waiting weeks or months for a response. I learned to keep pushing forward and not get discouraged by rejections or silence. Networking was a game-changer for me. In fact, I landed my current role at a startup through a contact I made at such event! After we met at this event I reached out to the CEO via email and after a few rounds of technical interviews I got a job. Those personal connections can accelerate your search in ways that blind applications can't. On the topic of applications - my advice is to start early and keep a steady pace. Don't make the mistake of trying to cram it all into a mad rush near graduation. I aimed to send a few well-crafted applications each week, which kept me motivated without being overwhelmed. Cast a wide net - alongside the tech giants, consider smaller startups where you can often take on more responsibility early. Another key strategy is to study the paths of those a few steps ahead of you. I found great value in following senior peers on platforms like LinkedIn, observing how they presented their work and what kinds of roles they moved into. You can learn a lot from example. In my case, the interview process for each company was different - but most included a mix of technical assessments (like coding challenges or system design questions), research presentations, and behavioral interviews probing my ability to collaborate and communicate. Rigorous preparation for each, balanced with self-care to manage stress, was essential. The key takeaway? Persevere, and trust the process. Keep refining your skills, expanding your network, and putting yourself out there. The right opportunity will come, often in the form or timeframe you least expect. And if you feel overwhelmed, I helped me a lot to spend some time outdoors, play some sports or cook with my friends!
What would you tell your younger you regarding building your current career?
As we've seen, a career in AI and robotics is an exciting journey full of continuous learning, impactful work, and the chance to shape the future. It's a path that rewards passion, perseverance, and a willingness to embrace the unexpected. For students just starting out, my biggest piece of advice is to dive in and get hands-on experience. Take on projects, join research labs, participate in competitions. The field moves fast, and the best way to keep pace is by doing. Don't be afraid to reach out to professionals for guidance - we're usually happy to mentor the next generation. Remember that while technical skills are crucial, they're only part of the equation. Develop your ability to communicate, collaborate, and continuously learn. These soft skills will serve you well no matter where your career takes you. Most importantly, stay true to what motivates you. The field of AI is vast, with opportunities to make a difference in countless domains. Whether your passion lies in healthcare, sustainability, space exploration, or something entirely different, there's a way to apply these technologies to drive meaningful impact. So stay curious, stay resilient, and stay excited. The future of AI and robotics is bright - and it's waiting for innovators like you to shape it. I can't wait to see what you'll build.
Final thoughts & tips
As we've seen, a career in AI and robotics is an exciting journey full of continuous learning, impactful work, and the chance to shape the future. It's a path that rewards passion, perseverance, and a willingness to embrace the unexpected. For students just starting out, my biggest piece of advice is to dive in and get hands-on experience. Take on projects, join research labs, participate in competitions. The field moves fast, and the best way to keep pace is by doing. Don't be afraid to reach out to professionals for guidance - we're usually happy to mentor the next generation. Remember that while technical skills are crucial, they're only part of the equation. Develop your ability to communicate, collaborate, and continuously learn. These soft skills will serve you well no matter where your career takes you. Most importantly, stay true to what motivates you. The field of AI is vast, with opportunities to make a difference in countless domains. Whether your passion lies in healthcare, sustainability, space exploration, or something entirely different, there's a way to apply these technologies to drive meaningful impact. So stay curious, stay resilient, and stay excited. The future of AI and robotics is bright - and it's waiting for innovators like you to shape it. I can't wait to see what you'll build.
Resources to dig in more
Fluent Python
If you're serious about Python programming, Fluent Python is must-read. This comprehensive guide goes beyond the basics to help you write concise, idiomatic, and effective Python code. It covers advanced topics like data structures, functions as objects, metaprogramming, and more. Whether you're a Python beginner or an experienced developer, this book will deepen your understanding of the language!
The Illustrated Transformer
This comprehensive blog post by Jay Alammar provides an in-depth, visual explanation of the Transformer architecture, which has revolutionized natural language processing and many other areas of machine learning. The article breaks down complex concepts into easily understandable illustrations and explanations, covering key components such as self-attention mechanisms, positional encoding, and the overall structure of encoder and decoder layers. It's an excellent resource for both beginners looking to understand the fundamentals of Transformers and more experienced practitioners seeking a clearer visualization of the model's inner workings. The step-by-step walkthrough of the Transformer's processing pipeline makes it an invaluable tool for grasping the intuition behind this powerful architecture.
Let's Reproduce GPT-2
This extensive video tutorial by Andrej Karpathy, a renowned AI researcher and former Director of AI at Tesla, offers a comprehensive, code-first approach to building Large Language Models (LLMs) from scratch. Over the course of five hours, Karpathy guides viewers through the entire process of creating a GPT (Generative Pre-trained Transformer) model, starting from a basic character-level language model and progressively adding complexity. The tutorial covers crucial concepts such as tokenization, self-attention mechanisms, and training techniques, all while implementing the model in PyTorch. Karpathy's clear explanations and live coding demonstrations make complex ideas accessible, providing invaluable insights into the inner workings of LLMs. This resource is particularly valuable for those who want to gain a deep, practical understanding of how modern language models are constructed and trained.
