How did AI Singapore use AIAP and 100E to build a “200” strong AI Engineering team with the Blue Ocean Strategy?
The management classic Blue Ocean Strategy book by Chan Kim & Renée Mauborgne describes the business environment with the terms ’red ocean’ and ‘blue ocean’.
Blue ocean strategy is the simultaneous pursuit of differentiation and low cost to open up a new market space and create new demand. It is about creating and capturing uncontested market space, making the competition irrelevant.
From https://www.blueoceanstrategy.com/
In short, avoid the red market and hunt in the blue market where your competitors are not playing.
When AI Singapore (AISG) started in June 2017, the task was to help 100 companies build their AI products and solutions through the 100Experiments (100E) programme. We started with four (4) AI engineers/data scientists with co-workers from my previous organisations I persuaded to join me.
We then put out a recruitment advertisement, and we received around 300 resumes, of which only ten were Singaporeans resumes. I only hired 1 Singaporean to join our team.
Being a government-funded programme and hosted by a local university, our salary structure and incentives could not match what the industry will pay. How can we build up an AI Engineering team to meet our 100E KPIs when we have big tech giants such as Google, Microsoft, Grab, Facebook, etc., looking to hire the exact talent profiles in Singapore?
We hunted in the blue ocean for these AI talents instead.
AI Apprenticeship Programme (AIAP)
AI Singapore launched the AI Apprenticeship Programme (AIAP)® in early 2018, where we actively hunted for AI talents with only two criteria:
- Must be a Singapore citizen
- Pass the AIAP technical test and interview
We deliberately avoided stating the academic qualifications the candidate needs to have, such as a computer science or engineering degree with a specialisation in AI or ML. Instead, we stated the skills and knowledge the candidate needed to have to join the apprenticeship programme. We searched for talents that we could nurture and develop in a short period of time.
Hence, we looked for passionate individuals who self-taught themselves AI/ML and Python (or R) programming and basic software engineering skills. They would have gone online and participated in various MOOCs or read up and practiced their AI/ML skills on platforms such as Kaggle. Only a handful of individuals receive formal education in AI/ML at the university level.
One of the biggest misconceptions that hiring managers have is that only Computer Science graduates can do AI and ML. Some of my best AI engineers and data scientists have degrees in economics, psychology, business, biology or industrial engineering.
We have found that the ability to learn fast, passion for solving data problems, and the love of working with data is key to being a good AI Engineer or Data Scientist.
Everyone can learn to program. Not everyone has a passion for data and using data to solve a business problem.
This strategy has worked out remarkably well for us as we have found passionate individuals from various disciplines who have decided to cross over to the AI/ML sector. They have taken it upon themselves to learn AI/ML at a deep level. Some have spent up to 3 years poring over their old university mathematics and statistics books.
Kevin – one of our senior candidates from AIAP batch #3, commented:
I borrowed my son’s CMU maths/stats and AI textbooks as I found online MOOC was not sufficient
Kevin Lee, AIAP 3
Watch the video interview with Kevin.
What they lacked was the opportunity to apply their knowledge to a real-world AI/ML business problem. The AIAP offered the participants the opportunity to work on a real-world industry AI project (100E is explained in the next section) or an internal AI Singapore project to build AI/MLOps platform tooling, AI solutions, and products (AI Bricks).
We carry out the project phase of 7 months in-house at AI Singapore premises. AI Singapore assigns AI engineers to work on the 100E project, collaborating with apprentices who are treated as full-time staff. They receive support from our AI/ML Dev Ops and Data Engineering team, as well as the various AI Heads who are experts in different AI/ML fields.
Internship vs Apprenticeship
We do not believe in sending “students” to an organization to intern in the internship model. In our AIAP apprenticeship model, we hot-house the apprentices. This accelerates their learning and skills. And by collaborating intimately with their peers daily and in an environment where everyone eats-drinks-sleeps AI/ML and deployment, real-world AI projects get delivered.
The end outcome is a Minimum Viable Product (MVP) co-created by AI Singapore and the sponsoring organisation. The combined team will typically work on 10-15 sprints over 7-months and deploy the model into production by the last month.
In Foster’s 2013 Data Science for Business book on page 321 – he says:
- Data science is a craft.
- Craft is learned by experience.
- The most effective learning path resembles the classic trade, where aspiring data scientists work as apprentices to masters.
- Apprentices become “journeyman.”
- Some become masters and take on apprentices.
The AI Apprenticeship Programme is one such learning path. AI mentors from AISG will guide the AI apprentices, with some of the mentors having recently completed the apprenticeship program themselves.
The 100 Experiment (100E)
The 100E programme is where AI Singapore matches and assembles a team of AI scientists, AI engineers, and AI apprentices to help an organisation solve their AI problem statement. The organisation brings the problem statement, dataset, domain expertise, engineering/IT resources and co-funding.
For a project to be undertaken by AI Singapore under the 100E programme, the company must meet the following criteria:
- be a Singapore registered entity with a local engineering/IT/AI team (or planned to have)
- have the dataset
- problem statement with ROI or strategic value to Singapore
AI Singapore has engaged more than 500 companies and 15,000 professionals through our outreach and industry programmes to date. We approved 66 100E proposals; completed and deployed 23 of the 100E into production.
AI Singapore has worked on projects in almost every primary industry cluster in Singapore.
Sompo Holdings Asia collaborated with AI Singapore to build an AI-powered fraudulent claims detection system for enhanced claims processes. Sompo received an AI Award for general insurance during Singapore Business Review’s recently concluded Technology Excellence Awards in 2020. Watch Sompo interview.
IBM Singapore collaborated with us to help them better classify the quality of different hardware products into different risk categories based on product return rates as well as predict future product return rates. The AI project involved using image processing, deep learning and time-series analytics. Engineers only need a few minutes to analyse the data, providing them with a glimpse into the future to pre-empt potential customer issues. The cost savings from this solution was also tripled! Watch IBM interview.
Conclusion
Many of the apprentices often have multiple offers even before they complete the programme. AI Singapore hires up to 30% of these apprentices, and the rest end up in organisations such as DSTA, Grab and DBS Bank today. Some have gone on to form their own AI startups.
Of the nearly 162 Singaporean AI apprentices trained to date, AI Singapore has retained around 50 of these apprentices as our AI Engineers in various AI Engineering roles such as AIML Ops, AI modelling, Data Engineering and AI Products teams. Many take on the role of mentors to train in-coming batches of apprentices. We also continue to hire from the industry experienced AI engineers, Singaporeans and foreigners, to supplement our home-grown team.
In recognition of the AI Apprenticeship Programme’s talent development innovation, IDC awarded AI Singapore the “Talent Accelerator for Singapore” in 2019.
Why AIAP works is that apprentices get to work on real-world AI projects – not toy and syntactic datasets and problem statements. They get to experience what it takes to deliver an AI project to an actual customer – including all the pains of working with some demanding customers, missing or limited datasets and changing user requirements.
One of our pioneer batch of AIAP graduates who now works for the defence industry said:
“There are many technical tutorials out there, but few offer the hands-on experience needed to address real-world problems, and that is one of the key differentiators of AISG’s AI Apprenticeship Programme (AIAP)®.”
Derek, AIAP 1
Growing our own timber – one apprentice at a time.
