From 'I don't know where to start' to a 30-day AI roadmap she built herself

The Person

Sireesha has worked the same client account in broadcast support for 11 to 12 years. Her clients use her company's equipment to broadcast for names like Sun TV and Tata Sky. Her day starts around 8am and her phone does not really stop. People know her name. People know she is available.

That is the problem.

The Before

She had access to Copilot at work and knew how to use it. What she did not have was a way to turn that into anything real.

Every YouTube video and Udemy course she found jumped straight into no-code, drag-and-drop tools without ever explaining what was happening underneath. She is a Python beginner, so none of it landed.

Her words: "It was complete chaos till I joined."

She found MasterDexter through a Substack post during a tea break.

The Constraint That Did Not Go Away

She did not get extra time once she joined. Office did not get quieter.

What changed is she started finding 30 to 60 minutes a day, sometimes by getting a separate laptop free of work restrictions, and used that time on a single thing: building a personal 30-day AI engineer roadmap with Claude, based on the inputs from the cohort.

"I dedicate maybe 1 hour a day, sometimes less. It's hectic in my office. But I still got the clarity I never had before. It's not like you need everything at once."

What She Actually Built

• Her own 30-day learning roadmap, iterated multiple times with Claude until it fit her pace and her domain knowledge. Added to her personal portfolio.
• Her first personal GitHub, separate from her work GitHub. She had never had one before.
• Small logging additions to the Neural Vault artifact and early Python scripts, pushed to that new GitHub.
• A blog post about a 48-hour hackathon her cohort ran, written for her portfolio first, with a cleaner Substack version planned next.

The Hardest Part

Artifact 4, Model Smith: the fine-tuning project. She had gotten through the first three artifacts fine but said this one operated on "another level."

She missed two live sessions because she was relocating, which compounded the difficulty.

Her way through it was asking Claude to explain model behavior specifically inside her own professional domain, since that domain expertise was the one thing she could lean on while the technical concepts were still new.

The After

She says she now understands roughly half of what is happening when the cohort discusses agent orchestration and agent workflows, up from understanding none of it before.

She describes the main shift as one of attitude as much as skill: "build something, do whatever, let it flow," instead of sitting with an idea and never starting.

She says she would now walk into an AI engineer interview and explain her work with confidence.

"Now I have the clarity I needed. If people are talking about something, I know what happens behind it. It's a structured, step by step pathway to become an AI engineer."

Where She Is Now

Still working through Cohort 1. Still in the same demanding job. Still building in small daily blocks instead of long stretches.

She has signed up to continue into Cohort 2.

Best for: Experienced professionals with almost no spare time, who need structure more than motivation.