As life moves forward day by day, sometimes it’s easy to forget how far we’ve come — personally, professionally, or both. Consider Simao Liu who came from Beijing to San Francisco just seven years ago for a master’s program in computer science. In her second year, she joined IBM as a QA intern and has spent the last six years making remarkable progress through the organization.
She now manages Unified Governance and Machine Learning for IBM Hybrid Cloud, having risen through the ranks as one of our best data scientists and developers. These days she works on innovative machine learning for data classification.
I hear you’re teaching machines to learn as they classify data. How easy is that?
It’s not! We’re attacking a non-classic ML problem. In particular, we’re working with natural-language processing but with very little labeled data, and there’s not a specific classification example that has enough data in it to use for training. For supervised learning, you need at least some labeled data to learn from, so we’re now using industry data from our industry model team, and we’re also using ML to learn from customer behavior.
So, let’s say an end user manually identifies a five-digit number as a zip code. Then what?
Right. When they do that, it’s a manual term assignment, and we use machine learning to infer the association rules and patterns. We then use the new knowledge to train our ML solution to improve future recommendations for automatic term matching.
It’s a challenge, but we found that it’s actually training our thinking in a machine learning way.
Sounds like you made a creative leap from pure machine learning to thinking in a way that’s informed by machine learning. Where did that creativity come from?
Oh, I think from playing games. You know when you play Grand Theft Autoor Assassin’s Creed, your character can go around the back of a house, climb a tree, go through the window, and assassinate someone from behind! What I mean is you take circuitous routes. I learned this when I was 3 or 4 - my father played games a lot, and I’d watch him every night. That’s how I got into computer science. I decided one night that I’d grow up and study the science that makes these games possible.
Do you miss China?
I miss my family and I miss the food. Coming from Hunan. You can’t get good Hunan food here.
Not even in San Francisco?
It’s not authentic. In Hunan — Chairman Mao was from Hunan, by the way —we have our own ways of cooking shrimp, and levels of spiciness I can’t even get to.
I remember seeing a food show where red peppers were steamed in a pot, then if you drink the water, your tongue becomes numb.
You’re making me homesick. The transition from Beijing though, that’s just a distant memory. It was seven years ago.
You came to get a masters at USF?
I did. Those two years were the best education I've had. The way the teachers designed the projects was completely different from education in China.
Not necessarily, but more open-ended. They designed projects to encourage us to apply the techniques and skills we learned from the courses. In China, we learned at a more theoretical level, with fewer opportunities to apply the skills.
Has seven years in Silicon Valley changed you?
Oh yes. Seven years ago, I was an intern. I couldn’t even speak up in meetings. Now, I lead a project and manage a team. I wouldn’t say, “Kudos to me,” but I grew so much, largely because my manager trusted me. A lot of people wouldn’t have thought I could lead a project, but he did, or if he didn’t - he took the chance. Now, I’ve improved a product and process end-to-end, because years ago, I was entrusted.
Home Town: Yiyang, China
Currently Working On: Implementing Machine Learning Solutions for Unified Governance and Integration
Currently Reading: Ship of Theseusby JJ Abrams
Favorite Programming Language: Python
Top 5 Hunan Dishes: (Best Translations I Can Find)
1. Spicy Crayfish/Crab
2. Chopped bell pepper fish head
3. Shredded Pork with Vegetables
4. Hunan-style barbecue
5. Spicy beef rice noodle
Dinesh Nirmal – Vice President, IBM Analytics Development
Follow me on Twitter: @dineshknirmal