初生之犢的華麗冒險 「來美工作後，才知道我沒有不夠好。」
陳知寧 （Jessica Chen）
李遠山總是謙稱自己「誤入高手村」，好似能夠在頂尖的大數據內容推薦公司Taboola一展長才，是因為單純的幸運。
李遠山是東華應數所碩士，博士時期「半路出家」在中央資工鑽研機器學習。原本的未來藍圖很直觀，完成學業便在台灣謀求教職，按部就班的進行資歷累積，未曾打算拓展國際職涯，亦沒想過會在異地一鳴驚人。
申請上科技部「博士創新之星LEAP計畫」以後，李以初生之犢的狀態來到洛杉磯，開始第一份正式工作。面對充斥著母語菁英的環境，李遠山並不怯場，他做好眼前的事情而不多想：「無論結局如何，在這裡都是很好的自我增值機會。」事實上，李的穩定紮實並非盲目樂觀，他才到Taboola第四個月，便習得如何處理不熟悉的巨量資料，並結合個人的深度學習專長，利用BERT以遷移學習的方式，解決了先前公司以色列團隊困擾已久的多語系自然語言處理問題。這項突破讓後續的產品開發有了巨大躍進，目前公司正陸續發展超過百種的語言版本。而李遠山提出的方法與嫻熟的操作，屢屢被首席工程師Jun Xu盛讚，認為他為公司創造了重要的貢獻。
歸納李遠山迅速適應的原因，除了公司給予的獨立發揮空間、精銳而緊密的團隊，LEAP計畫建構的人際網絡，亦是重要的助力。李的主管巫孟倫是第一屆LEAP的學員，在他摸索大數據處理時給了大量輔導與建議。李表示，LEAP建立了一個強大的支持群體，讓成員能夠心無旁鶩的投入，也讓他接觸到很多同鄉的佼佼者。
相較於在台時的不確定感、就業環境的限制，李出國工作後，重新校正了對自己的認知：「BERT的成功對我來說是個很好的起頭，表示我會的東西有用。」李遠山提到，由於原先數學系的訓練，他對理論的掌握高於coding的能力，剛開始很擔憂自己的所學無以致用，但最後發現，他的專長恰好與團隊互補，走過的路並無白費。「台灣的科技人才程度很好，操作面之外，大家普遍有碩士以上的資歷，擁有很好的研究能力，只是我們不知道如何展現自己最好的那一面。」他也表示，在美工作學習到的不只是技術，還有世界市場的思考，台灣企業多半把客群設定在國內，而不把市場面向全球；Taboola的產品目標是發展一百零四種語言，流通於全世界，這是整體視野的區別。出去一趟，才能真正辨識自己座落的位置，才有機會丈量能力的落差與極限，「如果我的經驗有價值，想跟剛畢業的人說，你們擁有可能性，你們值得一次脫胎換骨。」
Yuan-Shan Lee has always been very humble. During the interview, he smiled timidly and said: " I am a rookie among a bunch of masters ", as if it comes out of pure luck that he is able to show his talents in one of the top digital advertising platforms called Taboola. Taboola is a company that uses big data approaches to provide content recommendation toolboxes to publisher websites.
Yuan-Shan Lee received his master's degree from the Institute of Applied Mathematics at National Dong Hwa University, and switched to Computer Science & Information Engineering at National Central University for his doctoral degree. His original blueprint for the future is as straightforward as most fresh Ph.Ds: find a faculty position in Taiwan. He followed his predecessor’s advice to build his credentials. Lee never imagined that he could be a blockbuster in a foreign land, since he never even intended to expand his international career before.
After applying for the "LEAP Program" (launched by the Taiwanese Ministry of Science and Technology), Lee came to Los Angeles for his first formal work. Facing an environment full of native elites, Yuan-Shan Lee has no fear. He didn’t worry too much about competition. "No matter what is the result, it is a good opportunity for self-enhancement," he says. In fact, Lee's emotional stability is not based on blind optimism; he has been at Taboola for only four months, and he has already learned how to deal with a huge amount of unfamiliar information. Combined with his deep learning expertise by using the BERT transfer learning method to solve the problem of multilingual natural language processing, he has helped solve the problems that troubled the Israeli team in the company for quite some time. This breakthrough has made a significant contribution to product development. Yuan-Shan Lee’s performances are highly praised by the Principal Engineer, Jun Xu. Taboola is currently developing its product into more than one hundred different language versions.
Yuan-Shan Lee's rapid adaptation to different cultures and his success in solving difficult problems for the company has been aided not only by his work with what might be the top team in the world at Taboola, but also by the help of the LEAP Program which creates an excellent networking for young people who come here from Taiwan. Meng-Lun Wu, the director of Yuan-Shan Lee’ section, was the first LEAP fellow. He gave valuable advice to Lee when he explored big data processing. Lee said that LEAP encourages members to meet each other and build a strong supportive system among them.
Compared with his own uncertainty about his future career due to the perceived restrictive employment environment in Taiwan, Lee recalibrated his self-understanding after he went abroad, "The success of BERT is a good starting point, indicating that my expertise can be useful. " Because of his training, Yuan-Shan Lee is good at understanding theory better than coding. At first, Lee was worried that his specialty might be useless, but he finally found that his expertise was just complementary to the team. The learning process he has experienced is not in vain. "Taiwanese have outstanding scientific and technological talents. Most of us are highly educated with excellent research abilities. It’s a pity that we don’t know how to display the best part of us. " He also mentions that working in the US not only enhances his high-tech skills, but also builds a mindset of world markets. Taiwanese companies mostly target customers within the country instead of thinking about global markets. In contrast, the goal of Taboola is to develop 104 language products and distribute them all over the world. It is the difference between a Taiwanese vision and a global vision.
"If my experience is valuable, I would like to share with students who are just fresh out of university that you deserve the chance to go abroad and to be reborn as a new person," Yuan-Shan Lee concluded, with his timid smile.
更多精采內容，請登入駐洛杉磯科技組臉書：
https://www.facebook.com/tecolasd/
