Data engineer vs data analyst vs data scientist

Three job posts, one confused student

What do the three roles do?

Words you'll meet in this lesson

The three roles at a glance

The three roles at three levels

Side by side, and the roles in between

Roles in between (Advanced)

Indicative salary ranges in India (2026)

Indicative ranges compiled from salary guides published in 2026 that draw on job portals and salary aggregators (such as Glassdoor India and AmbitionBox). Real offers vary widely by city, company type and skills: product companies and global capability centres often pay far more than these ranges, and specialised skills (cloud, streaming, ML engineering) raise pay further. Always check current listings before deciding.

One business question, three roles

1. The question

   Monday morning, the city manager asks: “Why did Pune orders drop 12% last Friday, and will it happen again?”

   A vague question. Each role turns part of it into precise work.

2. Engineer: can we trust the data?

   SELECT date, COUNT(*) AS orders, MAX(loaded_at) AS last_load
   FROM silver.orders WHERE city = 'Pune' AND date >= '2026-10-13'
   GROUP BY date ORDER BY date;

   First, rule out a broken pipeline: were all of Friday's files loaded on time? They were, so the drop is real, not a data problem. Engineers save teams from analysing missing data.

3. Analyst: what happened and where?

   SELECT area, SUM(CASE WHEN date = '2026-10-17' THEN 1 ELSE 0 END) AS fri,
          SUM(CASE WHEN date = '2026-10-10' THEN 1 ELSE 0 END) AS last_fri
   FROM silver.orders WHERE city = 'Pune' GROUP BY area ORDER BY fri - last_fri;

   The drop is concentrated in two areas, during 7–10 pm, when heavy rain was reported. The analyst shows it on a dashboard and explains it in two sentences to the manager.

4. Scientist: will it happen again?

   from sklearn.ensemble import GradientBoostingRegressor
   model = GradientBoostingRegressor().fit(X_train, y_train)   # features: hour, area, rain_mm, match_day …
   print(model.score(X_test, y_test))

   A demand model using weather, time and area predicts rainy-evening drops. The scientist also designs an A/B test: do higher rider incentives on rainy evenings recover the lost orders?

5. Engineer again: make it run every day

   The model needs live rain data, so the engineer adds a weather feed to the pipeline and schedules the model's daily features.

   This handover from notebook to production is where ML engineers and data engineers work closely with scientists. The result: on rainy evenings, rider incentives switch on automatically.

A Monday at TiffinGo: three screens, one platform

Step through six moments of the day. Left: whose screen we are looking at. Right: which part of the data platform they are working on.

Careers in motion: three real-time scenarios

Watch the three roles respond to a live incident, see how careers grow over ten years, and (advanced) compare the skill profile each role needs.

Diwali night: the dashboard shows zero orders

20:47 on the busiest night of the year. The live orders dashboard suddenly drops to zero. Is the business down, or is the data down? Watch who does what, minute by minute.

Ten years in fast-forward: how the three careers grow

Riya, Kabir and Arnav start in 2026 as a junior analyst, engineer and scientist. Watch typical paths over ten years, including the switches people commonly make.

Skill profiles: what each role needs most

Seven skills, scored from 1 (basic) to 5 (deep), for a typical mid-level person in each role. Watch the three profiles appear, then compare them to your own.

Where these roles work

Role boundaries differ between companies; small startups often expect one person to do parts of all three.

Which role fits you? A tiny quiz engine

Each statement adds points to one or more roles. Change the answers to match yourself (True if the statement describes you) and run it. Then try the challenges: classify real job-description lines, find your skill gap, and plan a team.

statements = {
    "I enjoy building things that run automatically every night": {"engineer": 2},
    "I like explaining numbers to people in simple words":        {"analyst": 2},
    "Statistics and probability interest me":                     {"scientist": 2, "analyst": 1},
    "I like fixing things when they break":                       {"engineer": 2},
    "I enjoy making charts and dashboards":                       {"analyst": 2},
    "I want to predict the future from past data":                {"scientist": 2},
    "Linux, servers and cloud sound exciting":                    {"engineer": 2},
    "I like running experiments to test ideas":                   {"scientist": 2},
}
answers = [True, False, False, True, False, True, True, False]   # change these!

scores = {"engineer": 0, "analyst": 0, "scientist": 0}
for (statement, points), yes in zip(statements.items(), answers):
    if yes:
        for role, p in points.items():
            scores[role] += p

for role, s in sorted(scores.items(), key=lambda x: -x[1]):
    print(f"{role:9} {'#' * s} {s}")
print("Best match today:", max(scores, key=scores.get))

• statements = {...}A dictionary of dictionaries: each statement maps to the roles it points towards and how strongly.
• zip(statements.items(), answers)zip pairs each statement with your True/False answer, in order.
• '#' * sMultiplying a string repeats it: a quick text bar chart.
• max(scores, key=scores.get)The role with the highest score. A match today is not destiny; skills change with learning.

Run it on your PC: match your skills to job descriptions

Put your real skills in my_skills, paste real job descriptions you find online into job_descriptions, and run. The program shows how well you match each one and which skills to learn next. Save it as jd_matcher.py.

# Compare YOUR skills with three typical job descriptions and see the gap.
my_skills = {"python", "sql", "excel", "statistics", "power bi", "git"}

job_descriptions = {
    "Data Analyst": """Strong SQL and Excel. Build dashboards in Power BI or Tableau.
        Basic statistics. Communicate insights to business teams. Python is a plus.""",
    "Data Engineer": """Python and SQL. Build batch and streaming pipelines with Spark and Kafka.
        Orchestrate with Airflow. Cloud (AWS or Azure), Linux, Git, data modelling.""",
    "Data Scientist": """Python, SQL, statistics and machine learning (scikit-learn).
        Feature engineering, experiment design, model evaluation. Git; Spark is a plus.""",
}
vocabulary = ["python", "sql", "excel", "power bi", "tableau", "statistics", "spark", "kafka",
              "airflow", "cloud", "linux", "git", "machine learning", "data modelling",
              "communicate", "dashboards", "feature engineering", "experiment design"]

for role, text in job_descriptions.items():
    text = text.lower()
    wanted = {skill for skill in vocabulary if skill in text}
    have = wanted & my_skills
    missing = sorted(wanted - my_skills)
    score = round(len(have) / len(wanted) * 100)
    print(f"{role:15} match {score:>3}%  ({len(have)}/{len(wanted)} skills)")
    print(f"{'':15} learn next: {', '.join(missing[:4])}")

# In a terminal:  python jd_matcher.py

The role quiz, one line at a time

A shorter quiz with three statements. Watch the scores dictionary change only when an answer is True, and watch the inner loop add points to each role a statement mentions.

Worked examples: real problems, solved and traced in memory

Four career problems: recognising the work, reading job posts, closing a skill gap and planning a team. Each runs live and is traced.

fix a failed nightly load · build a sales dashboard · predict next week's demand · add a Kafka topic · explain a drop in KPIs · design an A/B test

engineer : fix a failed nightly load; add a Kafka topic
analyst  : build a sales dashboard; explain a drop in KPIs
scientist: predict next week's demand; design an A/B test

the job post text

{'engineer': 4, 'analyst': 2, 'scientist': 0}
Title says analyst; work says engineer

needed with priorities; you have python, sql, git

spark    priority 8
cloud    priority 8
airflow  priority 7
kafka    priority 6

pipelines 95 h, dashboards 60 h, models 25 h

engineer : 2 full-time + 15 h left over
analyst  : 1 full-time + 20 h left over
scientist: 0 full-time + 25 h left over
Combined leftover hours: 60 -> about 1.5 more people

Check yourself

Quick revision

Interview and viva questions

Practice problems

Career-flavoured coding exercises. The last two are planning decisions a data team lead makes.

engineer, analyst, engineer, scientist, analyst, engineer

{'engineer': 3, 'analyst': 2, 'scientist': 1}

tasks = ["engineer", "analyst", "engineer", "scientist", "analyst", "engineer"]
counts = {}
for r in tasks:
    counts[r] = counts.get(r, 0) + 1
print(counts)

analyst: sql, excel, power bi · engineer: sql, python, spark

Common: {'sql'}

analyst = {"sql", "excel", "power bi"}
engineer = {"sql", "python", "spark"}
print("Common:", analyst & engineer)

5 of 8

62 % match

print(round(5 / 8 * 100), "% match")

needs: python, sql, statistics, machine learning · have: python, sql

['machine learning', 'statistics']

needs = {"python", "sql", "statistics", "machine learning"}
have = {"python", "sql"}
print(sorted(needs - have))

title analyst · counts engineer 4, analyst 1, scientist 0

Real role: engineer

title = "analyst"
counts = {"engineer": 4, "analyst": 1, "scientist": 0}
real = max(counts, key=counts.get)
print("Matches title" if real == title else f"Real role: {real}")

you: sql 4, python 3, stats 4, cloud 1

scientist 11
analyst 10
engineer 9

me = {"sql": 4, "python": 3, "stats": 4, "cloud": 1}
roles = {"engineer": {"sql": 4, "python": 4, "stats": 1, "cloud": 4},
         "analyst": {"sql": 5, "python": 2, "stats": 3, "cloud": 1},
         "scientist": {"sql": 3, "python": 5, "stats": 5, "cloud": 2}}
fit = {r: sum(min(me[s], need) for s, need in req.items()) for r, req in roles.items()}
for r, f in sorted(fit.items(), key=lambda x: -x[1]):
    print(r, f)

spark 3, airflow 2, cloud 3, kafka 2 (weeks at 10 h)

6.7 weeks at 15 hours a week

plan = {"spark": 3, "airflow": 2, "cloud": 3, "kafka": 2}
hours_needed = sum(plan.values()) * 10
print(round(hours_needed / 15, 1), "weeks at 15 hours a week")

dashboards 30, pipelines 25, models 5

Hire first: analyst

needs = {"analyst": 30, "engineer": 25, "scientist": 5}
print("Hire first:", max(needs, key=needs.get))

2 · 2 · 1

Monthly team cost: Rs 410000

cost = {"analyst": 60000, "engineer": 90000, "scientist": 110000}
team = {"analyst": 2, "engineer": 2, "scientist": 1}
print("Monthly team cost: Rs", sum(cost[r] * n for r, n in team.items()))

Real-world scenario questions (asked in company interviews)

Career and teamwork questions asked in data interviews. Decide your answer first, then open the model answer.