// module: testimonials
Collected feedback from engineers who have completed one or more of the Nullabyte tracks.
← Back to Home3+
years running cohorts
180+
completion records issued
4.7
average end-of-track rating
92%
rate tutor feedback as useful
// module: reviews
Farrukh Harun
Backend Developer · Kuala Lumpur
"Track 01 straightened out a lot of bad habits I had accumulated working alone. The containers module in particular — I had been treating Dockerfiles as magic until this course forced me to understand what each layer was actually doing. The tutor feedback on my final project was specific enough to be genuinely useful."
Track 01 · June 2025
Nurul Zahira
Data Engineer · Penang
"The serving track was harder than I expected, which is probably a good sign. The cost accounting module was new territory for me — I had no idea how to think about inference cost per request before this. The load-testing exercises were where things clicked. I did wish there was more time on the caching section, but I understand why the schedule is what it is."
Track 02 · June 2025
Razif Yusof
ML Engineer · Selangor
"I came in knowing how to train models but with almost no experience running them in production. By the third assessed deployment in Track 02, I was confident enough to set up monitoring on a side project without referring back to the course material constantly. That felt like a real shift."
Track 02 · May 2025
Amirah Wan
Software Developer · Johor Bahru
"The version control for datasets section in Track 01 answered something I had been confused about for two years. Nobody had ever explained to me that a dataset change is as significant as a code change from a reproducibility standpoint. Eight weeks later I had restructured how my team handles data in two projects."
Track 01 · July 2025
Sivakumar Rajan
Platform Engineer · Cyberjaya
"I signed up for the Capstone because I wanted to work on a real problem with proper oversight, not just follow another set of exercises. The fortnightly architecture reviews were the most valuable part — having someone with actual production experience point out where my design assumptions were going to cause trouble before I built anything saved a lot of time."
Track 03 · May 2025
Lim Mei Ling
Research Engineer · Penang
"The cohort channel was quieter than some online programmes but more substantive. Questions got answers from the tutor rather than from other learners guessing. The material felt current — the quantisation approaches covered in Track 02 matched what I was reading about in papers at the same time."
Track 02 · June 2025
// module: case-studies
A backend developer at a fintech company in KL had taken on ML work but found that experiments ran only on his own laptop. No colleague could reproduce a training run, and there was no clear path from a working notebook to something the team could operate.
Completed Track 01 over seven weeks while working full-time, focusing the exercises on packaging an existing pipeline his team was using. Followed immediately with Track 02, bringing the packaged model into a serving context with monitoring and load-testing on a staging environment.
The team adopted the containerised pipeline format for new projects. A serving setup built during Track 02 went into a test environment within two months of the track ending. The learner described it as the first time a model had left his machine in a form a second person could run.
"The exercises were hard enough that finishing them meant something." — Farrukh H.
A platform engineer at a logistics company wanted to operationalise a route-prediction model that had been sitting in a research notebook for over a year. The company had no internal ML infrastructure expertise, and the engineer had to build understanding and a production system at the same time.
Enrolled in the Capstone Residency using the route-prediction problem as the project. Weekly sessions with the mentor covered architecture decisions; the fortnightly reviews flagged three design assumptions early that would have caused serving problems under real traffic patterns.
A working inference server with monitoring and documentation was delivered at the end of the residency and demonstrated to the company's engineering team. The technical report served as internal documentation for the system. The model moved from notebook to a maintained service within the residency timeline.
"The architecture reviews were where I learned the most." — Sivakumar R.
A research engineer at a Penang-based NLP team had extensive model training experience but had never been responsible for deploying or operating a model. Her team was preparing to move their first product from prototype to a small production deployment and she needed to close the infrastructure gap quickly.
Entered Track 02 directly, having assessed the prerequisites and confirmed her containerisation background was sufficient. Completed all three assessed deployments and the load-testing exercises, focusing the final deployment on a text classification model similar to the team's product.
The quantisation techniques from the track reduced inference latency on the team's prototype by around 30% when applied to their model. The monitoring setup built in the third assessed deployment was adapted and deployed to the product environment. The learner led the internal knowledge-sharing session on the topic.
"The material matched what I was reading in papers — it felt current." — Lim M.L.
// module: contact
Phone
+60 4-608 3175Address
9 Lorong Abu Siti, 10400 George Town, Penang
Office Hours
Mon–Fri 9am–6pm
Sat 10am–2pm MYT
// module: cta
Write to us with a few lines about your background and which track you are considering. We will respond with a suggested starting point.
Get in Touch