Phase 0 · Foundations
👤 set name
0%
Aug 2026 → May 2027 · ~9 months · ~26 h/week

From college kid to internship-grade ML engineer

By May 2027 you can build ML things from scratch, explain them in interviews, and walk into internship applications with proof — not "I did courses" but "here are my projects, here's my GitHub, here's my math."

✍️ Set your WHY — one line you'll read on the days you don't want to show up
0%0 / 0 tasks done
01

Setup

one time, this week

Install: Python 3.12 (python.org) · VS Code + Python extension + GitLens · Git (git-scm.com)

Accounts: GitHub · Google Colab (free GPU — your training rig) · Kaggle · LeetCode

Rules: commit to GitHub every day you code · one course at a time, max 1 course + 1 project running

🎯 Commit daily = green graph = proof. Employers check.
02

Time template

same every week, all 9 months
Mon–Fri3h evening main track (e.g. 8–11pm)
Weekend5h main track, one morning block
Daily45–60 min DSA, before anything else
Sunday30 min review — GitHub, notes, plan week

Total: ~20h main + ~5–6h DSA = ~26h/week. A part-time job. Show up like it pays you.

03

This week

your live position + the weekly ritual

This week's tasks

0day streak · open this site daily
0tasks done in current month

DSA — one problem a day

0h / 20h

Weekly check-in · Sunday

12

The DSA track

5–6h/week · non-negotiable · all 9 months

Striver A2Z (takeuforward.org) in order: arrays (M1–2) → binary search (M3) → linked lists / stacks / queues (M4) → hashmaps / recursion (M5) → trees (M6) → graphs (M7+) → DP (Phase 1).

Ramp: 2–3 → 7 problems/week by Month 7. Rule: >45 min on a problem → look at the solution, understand it, re-solve tomorrow from memory.

Why it matters: DSA is the first filter at every company — including the labs. No DSA = no interview, no matter how good your projects are.

Striver A2ZNeetCode 150 (2nd pass, later)LeetCode

13

Pitfalls

what kills this plan

1. Tutorial hell — 50% watch / 50% build, always. Two weeks of pure watching = stop and build.

2. Skipping math — your edge. Wasting it caps you at "runs notebooks."

3. Copy-pasting code — type everything. Muscle memory is real.

4. Project hopping — ship it ugly, improve later. Shipped ugly > unshipped perfect.

5. Comparing to Twitter geniuses — your only competitor is last month's you.

6. DSA procrastination — the #1 reason good candidates fail interviews.

7. No GitHub commits — commit daily. Employers check.

8. "Learning" by scrolling — reels aren't learning. Hands on keyboard.

14

If you fall behind

every real plan breaks — here's the recovery playbook

1 week slipped — compress. Skip optional tasks, protect the building blocks. Don't skip the week's project time.

2–4 weeks slipped — shift, don't quit. Push the calendar back by the time you lost. Keep Month 5's flagship project sacred — it's your internship ticket.

A month or more — re-anchor. Message the agent (me). We re-plan together instead of you quietly quitting. Never ghost the plan — that's the only real failure.

🛟 The plan bends. It only breaks if you stop talking.
15

Definition of done

May 2027 · tick them as they happen
✅ Phase 0 complete = you are internship-grade. Phase 1 (internship + LLM engineering depth) starts June 2027.
Progress saved in this browser · build in public