99%, Then 100%: What I Learned Training Pulchowk Students for AWS
Numbers like "100% pass rate" can come across as a flex, and that's not my intent here. A lot of students and professionals are sitting on the fence right now about whether to get into cloud, DevOps, AI, ML, or data — wondering if it's too late, too saturated, too technical. This post is for them. The Pulchowk story is just the proof.
Highlights, if you’re short on time
- I personally handled the AWS training at IOE Pulchowk Campus across two batches. The first batch closed with a 99% pass rate on the AWS Certified Solutions Architect — Associate (SAA-C03) exam, and the second batch went all the way to 100%
- Both batches were organized by Tekbay Academy / Tekbay Digital (formerly Adex International)
- Together with a fellow trainer, our combined AWS and ML programs have now crossed 400+ certified candidates
- Across all of Tekbay Academy / Tekbay Digital’s programs overall, the company-wide pass rate stands at 70%
- Four students across these programs have scored a perfect 1000/1000 on their exams
- The post closes with a section on what’s actually worth learning right now if you’re starting (or restarting) a career in cloud, AI, ML, or data — written for students, career-switchers, and early-stage professionals
The Result
I personally handled two AWS batches at IOE Pulchowk Campus, both organized by Tekbay Academy / Tekbay Digital (formerly Adex International).
The first batch closed at a 99% pass rate on the AWS Certified Solutions Architect — Associate (SAA-C03) exam. The second batch went one step further and hit a full 100%. Every single student who sat for the exam passed it.
Going from 99% to 100% across two consecutive batches at the same campus tells me the structure behind the training is doing the work, not luck or a particularly easy cohort. So that’s what this post is actually about.
Why Most People Actually Fail This Exam
Before getting into what worked, it helps to talk about what doesn’t.
The AWS Solutions Architect Associate exam isn’t really a memory test. It’s not “name the service that stores objects” or “what does EC2 stand for.” Most candidates who fail this exam already know that stuff. Where they trip up is the scenario questions — the ones that give you a business situation and three services that all technically solve it, but only one of them is the right call once you factor in cost, latency, fault tolerance, or compliance.
That’s a different skill entirely. It’s not knowledge, it’s judgment. And judgment doesn’t come from reading a study guide twice. It comes from seeing enough real systems, or near-real systems, that the trade-offs start to feel obvious instead of abstract.
How the Pulchowk Sessions Were Structured
I built this batch around two things running in parallel, and I think this combination is the actual reason the numbers look the way they do.
1. Exam-domain depth, the way the exam actually tests it
We went through all four SAA-C03 domains — resilient architecture, high-performing architecture, secure applications, and cost-optimized architecture — but instead of teaching them as separate chapters, we kept circling back to the same scenario from different angles. Same hypothetical company, same growing problem, but viewed once through a cost lens, once through a security lens, once through a performance lens. That repetition with a twist is what makes the “why this service and not that one” instinct actually stick.
2. Real-world anchoring for every single service
For every AWS service we touched, I tried to tie it to something a student could actually picture happening at a real company. Why does a streaming platform scale aggressively right before a big match. Why does a bank treat encryption and key management like a non-negotiable. Why would a startup migrating off old physical servers reach for serverless instead of just renting bigger machines.
The moment a service stops being an item on a list and becomes “oh, that’s the thing that stops Netflix from crashing during a season finale,” retention changes completely.
Press enter or click to view image in full size
3. Hands-on labs that mirror what the exam screenshots actually look like
A good chunk of SAA-C03 questions show you a console configuration and ask what’s wrong with it, or what’s missing. You cannot memorize your way through that. You have to have built a VPC yourself, written an IAM policy yourself, configured a load balancer yourself, and broken it at least once. So labs weren’t optional extras in this batch — they were a core part of every single session.
Zooming Out: This Isn’t Just One Batch
Pulchowk is one story inside a much bigger pattern. Alongside a fellow trainer, our combined AWS and Machine Learning programs have now crossed 400+ certified candidates — across cloud architecture tracks, DevOps bootcamps, and the AWS Machine Learning — Specialty track.
The 70% company-wide number is honestly the more important one to sit with here, more than the headline 100%. Individual batches can run hot. A 70% pass rate sustained across an entire academy, across hundreds of students with different backgrounds and starting points, is a much better signal that the training model actually works, and that the 99% and 100% Pulchowk results were a strong cohort inside a system that was already performing well, not a one-off.
That perfect score number, four people out of everyone trained hitting 1000/1000, is the stat I keep coming back to personally. A perfect score doesn’t really reflect hard work alone. It means a concept sat so cleanly in someone’s head that there was zero confusion left to lose marks over.
What This Actually Means for Anyone Thinking About a Career Switch Right Now
This is the part I think matters most, and it’s the reason I wanted this post to live on more than just one platform.
If you’re a student in Nepal, India, or anywhere else right now, wondering whether cloud, AI, ML, or data is “still worth it” in 2026 — here’s my honest take, from someone who’s spent the last couple of years training people who went on to actually get hired in this field.
Cloud isn’t a side skill anymore, it’s the foundation everything else sits on. Every AI model, every ML pipeline, every data platform you’ve heard of runs on cloud infrastructure underneath. You don’t need to choose between cloud and AI as separate career paths. Understanding cloud is what makes you useful once you eventually move into AI or ML roles, because someone has to deploy, scale, and secure those models in the real world.
The market isn’t looking for certificate collectors anymore. It’s looking for people who can reason through a problem. This is the single biggest thing I tell every new batch — a certification opens the door, but what keeps you in the room is being able to explain why you’d design a system one way over another. That’s true whether you’re going into Solutions Architecture, DevOps, MLOps, or data engineering.
A huge number of students I’ve trained came in with zero cloud background, just core CS fundamentals from their engineering degree. That’s enough of a starting point. You don’t need a head start to begin, you need consistency over a few months and a structured path instead of bouncing between random YouTube tutorials.
And if you’re aiming to work at or build a startup, the most valuable people in the room are usually the ones who understand how cloud infrastructure, data pipelines, and AI/ML systems connect to each other, not just one slice of it in isolation. That’s exactly why a lot of training I run now blends cloud fundamentals with the basics of data and ML instead of treating them as separate worlds.
If there’s one piece of advice I’d leave with anyone reading this and feeling stuck on where to start: pick one path, cloud, data, or ML, and go deep enough to actually build something real with it. Let the adjacent skills come naturally from there. Breadth without depth gets you nowhere in interviews. Depth in one area, with enough awareness of the rest, gets you hired.
Why I Keep Doing This
I’m an AWS Certified Solutions Architect — Professional, AWS Certified Educator, and on the Google Cloud side, a Professional DevOps Engineer and Professional Cloud Architect. I work as a Cloud and DevOps Engineer and Trainer at Tekbay Digital, formerly Adex International, and the Pulchowk program, along with our other AWS, DevOps, and ML tracks, is organized by Tekbay Academy / Tekbay Digital.
This isn’t a solo effort either. A fellow trainer at Tekbay has been working alongside me across the AWS and ML programs, and it’s that combined effort that pushed our certified candidate count past 400. Numbers like these come from a team putting in consistent work over a long stretch, not from any one person’s session.
The labs we run, the scenarios we discuss, the trade-offs we argue about in class, all of it comes from actual project work we’re doing outside the classroom too, not just exam-guide material.
What’s Next
We’re already lining up upcoming DevOps bootcamps and the next round of AWS certification tracks. If you’re a student, a working professional looking to pivot, or part of an institution that wants structured cloud, DevOps, or AI/ML training brought to your campus or company, I’d genuinely love to hear from you.
And to every student from this Pulchowk batch — thank you for the energy you brought into every single session, week after week. Watching all of you walk out certified is the whole reason I keep building these programs.
I’m Om Thakur, a Cloud and DevOps Engineer and Trainer at Tekbay Digital, formerly Adex International. I hold the AWS Certified Solutions Architect — Professional credential, I’m an AWS Certified Educator, and on the Google Cloud side I’m a Professional DevOps Engineer and Professional Cloud Architect. Over the past couple of years I’ve trained and mentored students and working professionals across Nepal and India in AWS, GCP, DevOps, and AI/ML fundamentals.
Enjoyed reading this content? Let the author know!
Your likes, comments, shares, and saves help creators reach more builders.
Loading recommendations
Loading article