In Hessen no pupil has to take Informatik, and the Gesellschaft für Informatik knows of no binding plan to change that. So in Darmstadt, coding and AI are learnt mostly outside the timetable. Below: the learners our English-language live classes suit, the Informatik Hessen schools do offer, which AI tools a learner in Germany may legally use and from what age, and a project on 65 years of Darmstadt temperatures where a model trained in the past quietly goes wrong in the present.
Live online since 2020 · 10,000+ students worldwide · models built and tested, taught in English
In short
Coding and AI classes in Darmstadt at Modern Age Coders are taught live online in English to children from 6, teenagers and adults: Coding for Kids with Scratch, AI and Machine Learning for Teens, and Data Science for adults. Each learner keeps one mentor, who works from India and fits pace and accent to the person; lessons run for fifteen to twenty people at once or privately. No Hessen pupil is required to take Informatik.
Starting points
Which coding or AI course suits a Darmstadt learner?
A child starts by making, a teenager by building models, an adult by working with real data. Three courses for those three starts.
Who takes coding and AI classes with us in Darmstadt?
Darmstadt calls itself a Wissenschaftsstadt, a city of science, and our learners here range from Grundschule children to researchers who want Python for their own data.
Learners and their usual first course (our recommendation)
Learner
First course
What they make in month one
Child, 6 to 8
Coding for Kids
A Scratch character that moves and talks
Child, 9 to 12
Python and AI for Kids
A turtle drawing and a number game
Pupil, 13 to 15
Python for Teens
A program that reads a file and answers a question
Pupil, 16 to 18
AI and Machine Learning for Teens
A first model tested on unseen data
Adult
Data Science
A clean dataset, a chart and a simple model
Every course is live, with one mentor who stays the same, and the learner always does the typing. A family can book different courses for different members, each at their own time.
Hessen schools
Does a Darmstadt school teach Informatik?
Some do, as an option. None has to. Everything in this section was checked against the GI monitor and the KMK calendar on 2 November 2026.
Optional, not required
The Informatik-Monitor 2026/27 states for Hessen that there is no compulsory Informatik across school types. At a Gymnasium, Informatik can be chosen as an elective from grade 7.
Looking at all sixteen Länder, the monitor puts Hessen among six without Informatik that is compulsory for all, and one of three, with Berlin and Brandenburg, where no binding plans to introduce it are known.
What that means at home
A Darmstadt child can finish school without ever having had a programming lesson, unless the school offers one and the child chooses it. Whether that happens depends on the school, the timetable and luck.
Our classes do not replace school Informatik and award no marks. They give a learner who wants to code a regular, structured place to do it, whatever the school offers.
When Hessen schools close in 2026/27
Break
From and to
Herbstferien
5 to 17 October 2026
Weihnachtsferien
23 December 2026 to 12 January 2027
Osterferien
22 March to 2 April 2027
Sommerferien
28 June to 6 August 2027
Informatik rules: GI Informatik-Monitor 2026/27 (Hessen and the Länder comparison). Holiday dates: KMK calendar for 2026/27, version of 9 October 2025; Hessen lists no winter week this year. Modern Age Coders works independently of the GI, the KMK and the Hessen Kultusministerium.
AI tools and age
Which AI tools may a learner in Germany use, and from what age?
Parents ask this more than any other AI question. The answer is in each company's own terms, and for Germany several are stricter than the general rule.
Minimum ages in the providers' own terms, as checked in our AI tools age audit
Tool
Minimum age
Note for Germany
ChatGPT (OpenAI)
13
Under 18 needs a parent's or guardian's permission
Google Gemini, personal account
13 generally
16 in Germany
Microsoft Copilot
13 generally
18 in Germany
Claude (Anthropic)
18
18 everywhere, or higher where local law requires
Source: our AI tools age guide, which quotes each provider's terms: OpenAI terms of use, Google Gemini help, Microsoft Copilot support and Anthropic consumer terms. Terms change; check the provider before signing a child up. This is not legal advice.
In our classes no one under 18 is asked to open any of these accounts. Children and teenagers learn how AI works by building small models in Python, which needs no account at all. Adults may use the tools in class, and the adult courses teach how to check what they produce.
Taught in English
What is it like to learn coding and AI in English from a mentor in India?
Every class is live and taught in English only. The mentor adapts to the learner, not the other way round.
No German lessons
We teach in English, and there is no German-language version of any course. We say it plainly here so that no family discovers it after booking.
Adapted to each learner
A mentor slows down for a seven year old and speeds up for a researcher. Accent, vocabulary and pace are matched to the person, and anything unclear is explained a second way.
The field speaks English
Programming languages, AI libraries and most research papers use English. Learning in English means reading the documentation of a new tool without a translation step.
Because the mentors work from India over live video, Darmstadt afternoons and early evenings are the usual class times. The same mentor teaches a learner for the whole course.
A real Darmstadt project
Does a model trained on Darmstadt's past still fit today? A lesson in data drift
The Wissenschaftsstadt Darmstadt publishes its average temperature for every month since January 1961. Teenage and adult learners work through it to see one of the commonest ways AI goes wrong: the model is fine, but the world it learnt from has moved on.
The data
The city's dataset Monatliche Durchschnittstemperatur ab 1961 gives the monthly mean temperature in degrees Celsius, based on measurements by the city and the Deutscher Wetterdienst. Reuse is allowed with credit under the Datenlizenz Deutschland Namensnennung, version 2.0. Our copy, fetched on 3 November 2026, holds 781 months from January 1961 to January 2026.
The city notes that the station moved several times, from the Botanischer Garten in the early 1960s to the Dieburger Straße since August 1995. Moves like that are a second, quieter kind of drift, and we tell learners about it.
The model
The simplest temperature model there is: for each month, predict the average of that month over a training period. Train it on 1961 to 1990 and it knows what a normal Darmstadt January or July looked like then.
We measure two things on test years. The average size of the error, and the bias: whether the errors lean one way. A good model has a bias near zero.
Version 1: train on 1961 to 1990, test old and newPython
temps = {}
with open("darmstadt_monthly.csv", encoding="latin-1") as source:
header = source.readline()
for line in source:
cells = line.strip().split(";")
year = int(cells[0])
for month in range(1, 13):
if cells[month] != "":
temps[(year, month)] = float(cells[month].replace(",", "."))
def learn(first, last):
normal = {}
for month in range(1, 13):
values = [temps[(y, month)] for y in range(first, last + 1)]
normal[month] = sum(values) / len(values)
return normal
def test(normal, first, last):
errors = [temps[(y, m)] - normal[m] for y in range(first, last + 1) for m in range(1, 13)]
size = sum(abs(e) for e in errors) / len(errors)
bias = sum(errors) / len(errors)
return round(size, 2), round(bias, 2)
old = learn(1961, 1990)
print("Months in the file:", len(temps))
print("Trained 1961-1990, tested 1961-1990:", test(old, 1961, 1990))
print("Trained 1961-1990, tested 2016-2025:", test(old, 2016, 2025))
It prints Months in the file: 781, Trained 1961-1990, tested 1961-1990: (1.45, 0.0) and Trained 1961-1990, tested 2016-2025: (1.55, 1.13).
On its own training years, the model has no bias at all, which is guaranteed: an average has no lean on the data it came from. On 2016 to 2025 the bias is 1.13. On average every month of the last ten years was 1.13 degrees warmer than the 1961 to 1990 model expected. The code did not change. The world did.
Version 2: train on recent years, or retrain every yearPython
recent = learn(1986, 2015)
print("Trained 1986-2015, tested 2016-2025:", test(recent, 2016, 2025))
rolling = []
for y in range(2016, 2026):
normal = learn(y - 30, y - 1) # retrain every year on the 30 years before it
rolling += [temps[(y, m)] - normal[m] for m in range(1, 13)]
print("Retrained every year, tested 2016-2025:",
(round(sum(abs(e) for e in rolling) / len(rolling), 2), round(sum(rolling) / len(rolling), 2)))
Run after Version 1. It prints Trained 1986-2015, tested 2016-2025: (1.35, 0.78) and Retrained every year, tested 2016-2025: (1.31, 0.66).
Error and bias on 2016 to 2025, in degrees C (our calculation)
Model
Average error
Bias
Normals from 1961 to 1990
1.55
+1.13
Normals from 1986 to 2015
1.35
+0.78
Retrained every year on the previous 30 years
1.31
+0.66
Training on more recent years helps, and retraining every year helps a little more, but the bias never reaches zero, because the warming continued even inside each training window. Averaging the city's own annual values gives 9.99 degrees for 1961 to 1990 and 11.09 for 2016 to 2025. The warmest year in the file is 2023 at 11.7 degrees, the coldest 1963 at 8.0.
Temperatures: Wissenschaftsstadt Darmstadt on opendata.darmstadt.de (Monatliche Durchschnittstemperatur ab 1961), credited under Datenlizenz Deutschland Namensnennung 2.0, file of 3 November 2026. Training periods, errors, biases and averages are our own. The project is about how models age, not a climate analysis; the station moves listed by the city mean the series is not a single site.
Children, 9 to 12. Write down what you wore to school last November. Would last year's clothes list work for this November? That is a model and its test.
Teenagers. Change the test years to 1991 to 2000 and then 2006 to 2015. Plot the bias of the 1961 to 1990 model against the decade.
Adults. Try training windows of 10, 20 and 30 years with yearly retraining. Which window gives the smallest error, and why might a short window be noisier?
Everyone. Name one AI system you use whose world could drift. What would you test it on?
Times
When are classes in Darmstadt time?
Times are agreed with each learner. These examples show a Darmstadt time next to the mentor's clock.
Darmstadt slot against the clock in India
Darmstadt
India before the spring change
India after 28 March 2027
14:50
19:20
18:20
17:10
21:40
20:40
Saturday 09:45
14:15
13:15
Sunday 12:30
17:00
16:00
Germany is on MEZ (UTC+1) for the winter and MESZ (UTC+2) for the summer, under the Einheitenzeitgesetz; this winter began on 25 October 2026 and summer time returns on 28 March 2027. India keeps UTC+5:30 throughout, so the mentor is four and a half hours ahead now and three and a half from late March. None of these rows is a fixed class.
Learners in Griesheim, Weiterstadt, Pfungstadt, Mühltal or Dieburg join just the same. Booking shows every slot in your own clock.
Progress
What can a Darmstadt learner do at each stage?
Our own stages, shared by every age group; only the speed differs.
Coding and AI stages (ours)
Stage
The learner can
1. Command
Make a computer follow exact steps
2. Code
Write Python with loops, functions and lists
3. Data
Read and clean a real dataset
4. Model
Train a simple model and measure its error
5. Maintain
Test a model on new data and decide when to retrain
For parents
The drift project reaches stage 5. A teenager who has seen a model go stale knows that "the computer worked it out" is never the end of the story.
Our reports describe what was built and explained.
For learners
A model is only as current as its training data.
Always test on the most recent data you have.
More courses
Further courses Darmstadt learners choose
The same live, English-language format, one mentor, class or private.
A weather forecasting site with live conditions for any location.
by Krish
Fees
What classes cost for Darmstadt learners
One monthly fee buys eight live classes. It appears here in euro at the European Central Bank reference rate; the payment is taken in US dollars and your bank converts it, so the euro on your statement can be slightly different.
Google reviews from students and parents, word for word.
★★★★★
"The one step solution for my son. Modern Age Coders make learning coding so simple that kids love it. The teachers explain complex concepts clearly with practical exercises and interactive content."
Ria Mukherjee
Parent
★★★★★
"Modern Age Coders has been a game-changer for me. I struggled to grasp IT concepts and coding before joining, but their classes transformed everything. I can now confidently write complex programs with ease."
Samriddha Mondal
Student
★★★★★
"One of the most wonderful education centres out there. Education is not limited to school syllabus but focuses on skill development."
Vansh Agarwal
Student
★★★★★
"My child Dhairya is really enjoying the Modern Age Coders classes. This is his first online class and he eagerly looks forward to it. I can already see his improvement, and the teachers are very cooperative."
Sonam Oswal
Parent of Dhairya
★★★★★
"Modern Age Coders have wonderful teachers who teach in a clear, easy and practical way. The teacher boosts students' confidence and inspires them to learn without hesitation."
Sonu Goyal
Parent
★★★★★
"I highly recommend this computer coding class! The teachers are incredibly knowledgeable and passionate about coding."
Ritu Kedia
Parent
Questions from Darmstadt
Coding and AI classes in Darmstadt: frequent questions
Are the classes in German?
No. All classes are live and taught in English only, by mentors based in India. They match each learner's pace and accent, explain things a second way when needed, and keep language simple for children.
Is Informatik compulsory in Darmstadt schools?
No. The Gesellschaft für Informatik reports that Hessen has no compulsory Informatik across school types; at a Gymnasium it can be an elective from grade 7. No binding plans to change this are known.
From what age can my child use ChatGPT or Gemini?
According to the providers' terms, ChatGPT is from 13 with a parent's permission under 18, Gemini with a personal account from 16 in Germany, Microsoft Copilot from 18 in Germany, and Claude from 18. Our classes never require a child to open any of these accounts.
When are classes in Darmstadt time?
Weekdays or weekends, flexible: choose your batch, or for a very personalised plan choose 1-on-1 at your convenience, weekend mornings or weekdays. Afternoons suit pupils; early evenings and weekends suit most adults.
What is data drift, and why do you teach it?
It is when the data a model learnt from no longer matches the world it is used in. Our Darmstadt project shows it with 65 years of temperatures. Real AI systems drift too, so learners must know to test on recent data and retrain.
Can my child start AI without coding?
Young children start with AI literacy and Scratch, which need no typing. Real AI work needs Python, which most children start from about nine and teenagers can learn quickly.
Is the temperature project real Darmstadt data?
Yes. It uses the Wissenschaftsstadt Darmstadt dataset Monatliche Durchschnittstemperatur ab 1961, under the Datenlizenz Deutschland Namensnennung 2.0, downloaded on 3 November 2026. The models and figures are ours.
Do you teach adults who already code?
Yes. Developers and researchers often go straight to the AI and ML Masterclass, Data Science or Generative AI, usually one to one so the projects can use their own data.
How big is a class?
A class holds fifteen to twenty people. Children are put together by age and older learners by how far they have got. Private learners get two lessons a week with the mentor alone.
What currency will I pay in?
US dollars. We display the monthly fee in euro, worked out from the ECB reference rate; when the dollar charge reaches your account, your bank converts it, so a little variation is normal. You pay nothing for a first class.
Expect a short task and a conversation. While the learner works, the mentor notes their approach and their ease with English; you then get a recommended course and group.
Every class is online, live and in English, taught from India. For a fast reply, write to us on WhatsApp.
Want a class sooner? Book a priority demo
A priority demo is a full class of about 45 to 60 minutes. After the class the mentor sends a written skill report and a personal learning roadmap on WhatsApp. If you enrol, the demo fee is adjusted against your first month's fee. Share your demo Order ID with us on WhatsApp and we take it off.
