“Learning machine learning” became 210 logged hours
DK
Deepa Krishnan
ℹ️
Illustrative example — a written scenario showing how the tool is used, not a real customer testimonial.
Before
Deepa had been learning ML for two years. In practice she watched tutorials when she felt guilty, retained little, and could not have told an interviewer what she had actually built.
What they logged
- Every practice session with duration and type — course, project, reading
- A self-rating out of 5 at the end of each session
- Milestones: first model deployed, first Kaggle submission
The moment it clicked
Her ratings made it plain: tutorial sessions averaged 2 out of 5 for value, project sessions averaged 4.5. She had spent 80% of her hours on the lower-value activity. She flipped the ratio.
What changed
| Measure | Before | After 15 months | |
|---|---|---|---|
| Logged practice hours | unknown | → | 210 hours |
| Project vs tutorial time | 20 / 80 | → | 70 / 30 |
| Outcome | no portfolio | → | Senior Analyst, 40% raise |
How they keep track now
Weekly practice minutes plus the running average rating. A skill averaging under 3 for a fortnight means the method is wrong, not her aptitude.
“Two years of “learning” with nothing to show. Fifteen months of logging, and I could point at every hour.” — Deepa Krishnan, Coimbatore
🧠 Try the Skill Tracker
Free, no card. Your data stays yours.
Open Skill Tracker →
More Learning & Growth stories