How NumPy Skills Boost Your Data Engineer Job Search

Weeks 1-2 — Daily Schedule:
7:00 AM — Wake up, exercise (30 min)
8:00 AM — Review NumPy topics for the day
9:00 AM — Deep work (3 hours)
   - Week 1: ndarray, shapes, dtypes, indexing
   - Week 2: Broadcasting, ufuncs, aggregations
12:00 PM — Lunch break
1:00 PM — Practice problems (LeetCode / HackerRank)
2:00 PM — Deep work (2 hours)
5:00 PM — Review progress, plan tomorrow
6:00 PM — Exercise, rest
8:00 PM — Read NumPy docs / blog posts (30 min)
10:00 PM — Wind down

Week 1 Review:
□ Can I create ndarrays of any shape and dtype?
□ Do I understand indexing and slicing?
□ Can I reshape and transpose arrays confidently?

Week 2 Review:
□ Do I understand broadcasting rules?
□ Can I use ufuncs and aggregations correctly?
□ Have I replaced at least 5 loops with vectorized code?
□ Do I know the difference between axis=0 and axis=1?

Weeks 3-4 — Daily Schedule:
7:00 AM — Wake up, exercise (30 min)
8:00 AM — Review pipeline goals for the day
9:00 AM — Deep work (3 hours)
   - Week 3: Load CSV, clean, reshape with Pandas + NumPy
   - Week 4: Vectorization, memory layout, benchmarks
12:00 PM — Lunch break
1:00 PM — Build the ETL mini-project
2:00 PM — Deep work (2 hours)
5:00 PM — Review progress, plan tomorrow
6:00 PM — Exercise, rest
8:00 PM — Read data engineering blogs (30 min)
10:00 PM — Wind down

Week 3 Review:
□ Does my ETL script load real CSV data?
□ Do I convert the numeric columns to NumPy arrays?
□ Have I handled missing values with np.isnan / np.nan_to_num?

Week 4 Review:
□ Have I replaced slow loops with vectorized operations?
□ Do I understand memory layout (C-order vs F-order)?
□ Can I benchmark a vectorized vs loop implementation?

Weeks 5-6 — Daily Schedule:
7:00 AM — Wake up, exercise (30 min)
8:00 AM — Review interview topics for the day
9:00 AM — Deep work (3 hours)
   - Week 5: Array coding problems (LeetCode / StrataScratch)
   - Week 6: NumPy in ML pipelines, data infra design
12:00 PM — Lunch break
1:00 PM — Mock interviews / pair coding
2:00 PM — Deep work (2 hours)
5:00 PM — Review progress, plan tomorrow
6:00 PM — Exercise, rest
8:00 PM — Behavioral prep (30 min)
10:00 PM — Wind down

Week 5 Review:
□ Have I solved 50 array problems?
□ Can I explain broadcasting in plain English?
□ Do I know how to convert between Pandas and NumPy?

Week 6 Review:
□ Have I done 5 mock interviews?
□ Can I describe a NumPy-powered pipeline I built?
□ Do I have a clear project story (problem → solution → impact)?
□ Have I prepared questions to ask the interviewer?

Array Playground — practice file:
import numpy as np

# Create a 2D array of shape (3, 4)
a = np.arange(12).reshape(3, 4)

# Broadcasting with a scalar
b = a * 2

# Column-wise mean
col_mean = a.mean(axis=0)

# Boolean mask
mask = a > 5
filtered = a[mask]

print(a)
print("Column means:", col_mean)
print("Filtered:", filtered)

ETL Pipeline — mini-project:
import numpy as np
import pandas as pd

# Load raw data
raw = pd.read_csv("events.csv")

# Convert numeric columns to NumPy for fast processing
arr = raw[["user_id", "amount"]].to_numpy()

# Vectorized cleaning
arr = arr[~np.isnan(arr[:, 1])]
arr[:, 1] = np.clip(arr[:, 1], 0, None)

# Aggregate per user
users, counts = np.unique(arr[:, 0], return_counts=True)
summary = dict(zip(users, counts))
print(summary)

Benchmarking — proof of impact:
import numpy as np
import time

n = 1_000_000
data = np.random.rand(n)

# Vectorized
start = time.perf_counter()
s1 = np.sum(data * 2)
vec_time = time.perf_counter() - start

# Loop
start = time.perf_counter()
s2 = 0
for x in data:
    s2 += x * 2
loop_time = time.perf_counter() - start

print(f"Vectorized: {vec_time:.4f}s | Loop: {loop_time:.4f}s")
print(f"Speedup: {loop_time / vec_time:.1f}x")