United States · Live online · Ages 6 to 67
Washington has issued an executive order telling federal agencies to build AI literacy for school students. Forty six of the fifty state computer science standards still do not name artificial intelligence anywhere inside them. American families stand in the gap between those two facts, and that gap gets closed privately or not at all. We teach the closing: Python, the statistics underneath it, and models a student trains, scores and publishes at an address anyone can open.
Teaching since 2020 · 10,000+ students across 25+ countries · no recordings sold as classes
In short
Modern Age Coders teaches live online AI and machine learning to learners across the United States between the ages of 6 and 67, in groups of five to eight or privately. The syllabus runs from Python and the mathematics underneath it to models the student trains, evaluates and publishes at a working link. Weekend cohorts serve both coasts and a weekday cohort runs on Pacific time. Group tuition is $40 a month, one to one teaching is $100 a month, and the first class costs nothing.
The catalogue
There is no fast lane and no slow lane here. There are three different subjects sharing a name on a timetable. A child learns what it even means for a rule to be discovered from examples. A teenager learns to hold data back and then state an error out loud. An adult learns to deploy something and watch it drift. Every card opens a week by week outline.
US / KIDS / 01
Typed code from the first session rather than coloured blocks. Python, turtle graphics and small games, then the sentence the whole subject hangs from: some rules are discovered from examples instead of written down by a person.
Open the syllabusUS / KIDS / 02
What sits behind a chatbot reply, why confidence tells you nothing about accuracy, and how to test a statement against a source before handing it in. At this age that habit outlasts any tool.
Open the syllabusUS / KIDS / 03
A child gives an AI tool an instruction and something half working comes back. Repairing the broken half is the course. Writing instructions takes an afternoon. Reading output sceptically takes a term.
Open the syllabusUS / TEEN / 01
The main track on this page. Python, the mathematics the subject requires, classical algorithms, neural networks, computer vision and language models, every stage ending in something trained and scored instead of described.
Open the syllabusUS / TEEN / 02
Loading, cleaning, joining and charting, where most working hours in this field go. Teenagers learn to catch the mislabelled column, the duplicated record and the missing value coded as a zero.
Open the syllabusUS / TEEN / 03
Two years from a first print statement through files, APIs, testing and object oriented design. Take this first if your teenager has never programmed; the machine learning track assumes the fluency it builds.
Open the syllabusUS / TEEN / 04
Teenagers ship web and Python projects with an AI assistant beside them, then sit through a review of what it produced. The deliverable is a working URL they can also explain line by line.
Open the syllabusUS / PRO / 01
Supervised and unsupervised learning, deep networks, validation that survives an outsider reading it, and the deployment stage short programmes omit precisely because it exposes a weak model.
Open the syllabusUS / PRO / 02
How large language models are pre trained, fine tuned and grounded against a private document set, and how an agent is fenced in so that when it goes wrong it goes wrong visibly.
Open the syllabusUS / PRO / 03
For American professionals who want the applied half first: machine learning and language models pointed at the reports, spreadsheets and pipelines already on their desk.
Open the syllabusUS / PRO / 04
Statistics, SQL, feature construction, modelling and the writing up, ending with a portfolio built to be read by a hiring manager rather than marked against a rubric.
Open the syllabusUS / PRO / 05
The technical interview gate at every serious American employer, and the reason a pipeline that runs on fifty one rows still runs on fifty one million.
Open the syllabusUS / ALL / 06
Outside AI we teach mathematics from early number work to university level, plus web, app and game development and exam preparation, all live and on US timings.
Open all coursesWhat you actually build
Every market we teach gets a first project on public data from that country, and the United States has the richest federal open data anywhere. The source is the Energy Information Administration's Electric Power Monthly, Table 5.6.A, average price of electricity to ultimate customers by state, residential sector, from Form EIA-861M. The edition read here is labelled "Data for May 2026, Release Date: July 23, 2026", and its own footnote says the values are preliminary estimates based on a cutoff model sample. Students meet that footnote on day one, because a dataset honest about how provisional it is deserves quoting with the warning attached.
| State | May 2026 | May 2025 | Year on year |
|---|---|---|---|
| Hawaii | 52.00 | 41.03 | +26.7% |
| California | 33.25 | 33.29 | -0.1% |
| New York | 29.93 | 26.69 | +12.1% |
| Massachusetts | 28.82 | 29.90 | -3.6% |
| Connecticut | 27.37 | 31.59 | -13.4% |
| Illinois | 23.85 | 18.58 | +28.4% |
| New Jersey | 23.27 | 20.48 | +13.6% |
| Texas | 16.44 | 15.53 | +5.9% |
| Oregon | 16.27 | 15.91 | +2.3% |
| Florida | 15.17 | 14.97 | +1.3% |
| Washington | 14.95 | 13.65 | +9.5% |
| Wyoming | 14.80 | 13.82 | +7.1% |
| Tennessee | 14.47 | 13.73 | +5.4% |
| Louisiana | 14.15 | 13.26 | +6.7% |
| Missouri | 13.68 | 13.01 | +5.2% |
| North Dakota | 13.61 | 13.10 | +3.9% |
| Nebraska | 13.59 | 13.15 | +3.3% |
| Oklahoma | 13.38 | 12.94 | +3.4% |
| Utah | 12.96 | 12.60 | +2.9% |
| Idaho | 12.35 | 11.88 | +4.0% |
| United States | 18.44 | 17.37 | +6.2% |
The two price columns are published by the EIA. The year on year column is computed from them. The May 2026 column is preliminary and will be revised in a later edition, which is stated here for the same reason it is stated at the source.
Divide Hawaii by Idaho and you get 4.21. A twelve year old can hold that ratio in their head and a seventeen year old can turn it into a modelling problem. Why would the same commodity, delivered the same way, cost four times as much in one state as in another?
The target is not the price. It is the year on year percentage change in a state's residential price, and that switch is the most important decision in the project. Prices did not simply move upward everywhere either: Connecticut fell 13.4 percent while Illinois rose 28.4 percent, so any story beginning "energy got more expensive" is too coarse to use.
Features come from the EIA State Electricity Profiles, 2024 data released 10 November 2025, which publishes net summer capacity, net generation and total retail sales for every state. Students derive ratios: generation over retail sales says whether a state exports or imports power, and generation over capacity times hours gives a rough capacity factor.
A trap sits inside that second table and it is the likeliest way this project fails in the real world. The State Electricity Profiles also carry an average retail price column whose national figure is 12.68 cents per kilowatt-hour, because it averages across all sectors. The target here is the residential price, national figure 18.44 cents. A student who joins the two tables on a column called "price" has silently merged two metrics, and no error message will say so.
pandasscikit-learn matplotlibordinary least squares leave one out CVmean absolute error
Pull Table 5.6.A for all fifty states plus the District of Columbia, residential sector, current month and year-ago column. Type the twenty rows above in by hand first, then check the parse against the printed national figures of 18.44 and 17.37. If the totals do not reproduce, the parse is wrong. The data is not.
Put May 2026 on one axis and May 2025 on the other. The points sit almost exactly on the forty five degree line, and that chart kills the obvious framing before any modelling starts. That framing is to predict the May 2026 price with last year's price as a feature, watch R-squared land near 0.97 and declare victory. Nothing has been learned except that prices are sticky, because the answer was handed to the model in its own input.
Switch to year on year percent change. Predict that every state moved by the national average of positive 6.2 percent and record that baseline's mean absolute error. Any model that cannot beat the number is worthless, and most first attempts do not. Writing it down first stops a student quietly moving the goalposts later.
Derive ratios rather than merging raw numbers. Raw net generation only encodes how large a state is, and Texas will dominate every coefficient for no interesting reason. Check the price column trap before the merge, not after the results look strange.
Fit ordinary least squares and validate leave-one-out. Do not use an eighty twenty split: with fifty one rows a ten row test set makes the error swing wildly with the random seed, and a student who reruns it twice gets two different stories. Report the leave-one-out mean absolute error against the step three baseline.
Hawaii is a leverage point. Its island grid burns imported fuel oil and it sits nearly nineteen cents above the next highest state, so with at most fifty one rows that one observation can anchor a least squares fit by itself. Refit without it and record which coefficients changed sign and which changed magnitude. That comparison, not the accuracy score, is the deliverable.
American retail electricity prices are set by state public utility commissions through rate cases. No feature built from generation mix can see a regulatory decision, so part of the variation is structurally invisible here. Say so, and note that the May 2026 column is preliminary. Being able to write "here is the part my data cannot explain" is the skill this exercise exists to teach.
The distinction
We teach both, in that order, and we tell parents plainly which of the two holds its value. An interface gets redesigned every year. Probability and matrix algebra were not redesigned last year and will not be next year.
The distinction lands harder here than in almost any market we teach, because the federal instruction to teach AI and the state standards that would deliver it are not the same document and were not written by the same people. A directive exists at the top, forty six state standards are silent in the middle, and individual families make up the difference at the bottom.
Why now, in the United States
There is no funded national AI education programme here in the sense a ministry would recognise. There is a sequence of executive actions, and reading them in order explains the market. Executive Order 14179, "Removing Barriers to American Leadership in Artificial Intelligence", was signed in January 2025 and commissioned the action plan that followed.
The order that matters to a school is Executive Order 14277, "Advancing Artificial Intelligence Education for American Youth", signed 23 April 2025 and published in the Federal Register on 28 April 2025. It establishes the White House Task Force on AI Education, and directs it to produce plans for a Presidential Artificial Intelligence Challenge within 90 days, with agencies holding the Challenge no later than 12 months after those plans are submitted. The stated purpose is to encourage and highlight student and educator achievements in AI, and to bring government, academia, philanthropy and industry together. One qualification belongs beside it: the order carries no appropriated dollar figure. It is a directive instrument, and a directive does not by itself put a machine learning teacher in a class.
"Winning the Race: America's AI Action Plan" arrived on 23 July 2025 with more than 90 federal policy actions across three pillars: accelerating innovation, building American AI infrastructure, and leading in international diplomacy and security. It is led by the White House Office of Science and Technology Policy with the AI and Crypto Czar's office, and three further executive orders on infrastructure and technology export were signed the same day. It states no dollar figures, and its named high-demand occupations are electricians and HVAC technicians. That is a plan to staff data-centre construction, not to teach a fourteen year old how a gradient descends. The children's mandate lives in Executive Order 14277 and nowhere else.
Now the delivery side. Code.org's 2025 State of AI and CS Education report, data finalised 1 December 2025 and released 10 December 2025, found that 60 percent of US public high schools offer foundational computer science, that only 4 of the 50 states explicitly emphasise AI within their computer science standards, and that only 5 of the 50 fund AI and computer science professional development for teachers. No state requires both AI and computer science for graduation.
The population is large. The National Center for Education Statistics puts public school enrolment from prekindergarten through grade 12 at 49.6 million in fall 2022, projected to fall to 46.9 million by fall 2031, about 5 percent. That indicator was last updated in May 2024, so read it as scale rather than a current-year count. We print no market-size dollar estimate for American online AI or coding education, because every figure available traced back to a vendor press release rather than a statistical agency.
The Bureau of Labor Statistics Occupational Outlook Handbook is employment evidence we will stand behind. Data scientists had a median annual wage of $112,590 as of May 2024, with employment projected to grow 34 percent from 2024 to 2034 and about 23,400 openings each year across the decade. Computer and information research scientists had a median of $140,910 as of May 2024, with 20 percent projected growth and about 3,200 annual openings.
The infrastructure story is real and frequently oversold, so here is the careful version. The Stargate Project was announced at the White House in January 2025 by OpenAI, SoftBank and Oracle with Abu Dhabi's MGX, at $100 billion initial funding ramping toward $500 billion over four years, the flagship campus in Abilene, Texas on Oracle Cloud Infrastructure. Five further sites were announced on 23 and 24 September 2025: Shackelford County in Texas, Dona Ana County in New Mexico, Lordstown in Ohio, Milam County in Texas, and a Wisconsin site developed by Oracle with Vantage. Stargate is data-centre construction, and its direct hiring is electricians, welders and site operators. It shows where American AI compute is being built and why the industry is spreading inland. It is not a hiring pipeline for a teenager learning gradient descent. The BLS rows above are.
4 of 50
States whose computer science standards explicitly emphasise AI, per Code.org's 2025 report
23 April 2025
Executive Order 14277 directs federal agencies to build AI literacy for K-12 students
+34%
Projected growth in data scientist employment, 2024 to 2034, Bureau of Labor Statistics
4.21x
Hawaii's residential electricity price divided by Idaho's, the hook for the first project
School and university
Families arriving from other systems keep looking for one. The United States has no national curriculum and no national exam board. Education is controlled by the fifty states and roughly 13,000 local school districts, so any provider implying an American equivalent of CBSE or Cambridge is describing something that does not exist. What exists is a shared reference document, the CSTA K-12 Computer Science Standards, which states adapt. Twelve states require a computer science course to graduate: Alabama, Arkansas, Indiana, Louisiana, Nebraska, Nevada, North Carolina, North Dakota, Rhode Island, South Carolina, Tennessee and West Virginia, the last being the most recent addition.
The country codes programmes rather than exams. The National Center for Education Statistics maintains the Classification of Instructional Programs, in its fifth revision as CIP 2020, where 11.0101 is Computer and Information Sciences, General and 11.0102 is Artificial Intelligence, with secondary courses additionally coded under the NCES School Codes for the Exchange of Data. AI having its own CIP code is worth knowing: the federal statistical system already treats it as a distinct discipline where state standards do not.
Advanced Placement, run by the College Board, is the closest thing to a national qualification in this subject. AP Computer Science A is a programming course in Java. AP Computer Science Principles is broader and includes a through-course Create performance task submitted online rather than sat entirely as a written paper. Neither is a machine learning course, and both read better as prerequisites in spirit than as destinations.
For reference, the 2026 administration ran 4 to 8 May and 11 to 15 May 2026, with AP Computer Science Principles on Thursday 14 May 2026 and AP Computer Science A on Friday 15 May 2026, both at 12pm local, and the Create performance task deadline of 30 April 2026 at 11:59pm Eastern. The College Board has published 30 April 2027 for the next Create deadline, but had not published the full per-subject 2027 grid as of July 2026, typically releasing it around the preceding autumn, so we print no specific 2027 subject dates. The International Baccalaureate Diploma Programme, including Computer Science at Standard and Higher Level, is offered at many American high schools alongside or instead of AP, and our teenage track maps onto it the same way.
Carnegie Mellon University's Bachelor of Science in Artificial Intelligence, run from its School of Computer Science, is the reference undergraduate AI degree in the country. Its own admission page lists the required high school coursework: 4 years of English; 4 years of mathematics, which should include at least algebra, geometry, trigonometry, analytic geometry and elementary functions and preferably calculus, with advanced mathematics encouraged; 1 year of physics; 2 years of chemistry, biology or computer science; 2 years of a foreign language; and 3 electives.
Notice the shape. That is a coursework profile, not a threshold. The page publishes no minimum grade point average, no test-score cut-off and no acceptance rate for the major, so any page quoting one has invented it. What it does say is that four years of mathematics, ending as close to calculus as your school allows, is the thing being checked.
Two programmes come up in almost every conversation with an American adult. The University of Texas at Austin runs an online Master of Science in Artificial Intelligence of 30 credit hours, 3 hours of required courses plus 27 hours of electives across 10 courses, described by the university in its own words as an affordable, advanced degree priced at $10,000 plus fees, with international student fees and late registration fees possible. Fall applications open 15 December, priority 15 March, final 15 April; spring open 1 June, priority 1 August, final 1 September.
Georgia Tech's Online Master of Science in Computer Science carries a machine learning specialisation, is also 30 credit hours and typically takes three years. Its prospective-student FAQ states the GRE is not required and that the preferred qualification is an undergraduate degree in computer science or a related field, typically mathematics, computer engineering or electrical engineering, with a cumulative grade point average of 3.0 or higher, others evaluated case by case. Materials include a bachelor's degree from a regionally accredited institution, transcripts, three recommendation letters submitted online by the recommenders themselves, English proficiency documentation for international applicants, and a misconduct-history disclosure. Tuition is per credit hour at University System of Georgia rates, and because the FAQ prints no total, we do not quote one.
The teaching implication is the same at both ends of that ladder. Whether a student aims at Carnegie Mellon out of high school or at a graduate programme ten years later, the binding constraint is mathematical depth, meaning calculus and linear algebra, not the number of AI products they have used. That is why our teenage track teaches the mathematics inside the machine learning instead of parking it in a separate subject to be survived.
How the classes run
The contiguous United States spans Eastern to Pacific, and India does not observe daylight saving, so the gap between us moves twice a year. American daylight saving runs from the second Sunday in March to the first Sunday in November. In July 2026 Eastern Time is UTC-4 and India is 9 hours 30 minutes ahead of it; Pacific Time is UTC-7 and India is 12 hours 30 minutes ahead. From November to March those become 10 hours 30 and 13 hours 30. Timetables are rebuilt on both Sundays, or every American class silently slides by an hour.
Saturday 11:00am ET
8:00am Pacific, and 8:30pm Saturday in India. The flagship cohort. It works for both coasts and lands in an ordinary Indian evening, which is why we protect it.
Sunday 12:00pm ET
9:00am Pacific, and 9:30pm Sunday in India. The second weekend cohort, still comfortable at both ends of the call.
Tue and Thu 6:30pm PT
7:00am Wednesday and Friday in India, the next calendar day. Pacific coast only, because 6:30pm Pacific is already 9:30pm Eastern.
Ninety minutes on live video with a teacher who remembers what went wrong last week and opens with it. Both screens are shared, so an error gets interrupted while the student is still making it rather than annotated three days later. Batches hold five to eight students, or one. Between one week and the next there is a build rather than a worksheet, and questions in the gap go to that same teacher instead of a ticket queue staffed by someone who has never met the child.
The honest limit is worth stating once. A weekday evening class at 6:00pm to 8:00pm Eastern falls between 3:30am and 5:30am in India. No teacher on a normal schedule holds that slot week after week, so we do not sell it. We serve the United States fully at weekends, and on weekdays we serve the Pacific coast.
There is no national school calendar. Start and end dates are set by roughly 13,000 local districts, so they vary by weeks. New York City Public Schools, the largest district in the country, begins the 2026 to 2027 school year on Thursday 10 September 2026. Most districts run from mid-August or early September to late May or June. Thanksgiving is a fixed federal anchor on the fourth Thursday of November, which is 26 November 2026, and most districts close the Thursday and Friday. Winter break runs from shortly before Christmas to shortly after New Year.
AP season dominates late April into mid May, when attendance dips and demand for exam-focused help spikes. Spring break is a single week in March or April and is not synchronised nationally, which is why an American cohort loses different students in different weeks that month and why we do not pause a batch for it. September and January are the natural intake points, and June to August is the deepest window, because school is out and a weekday morning slot becomes workable.
Enrolment does not run in cohorts with a closing date. A student is placed into the next batch that fits their level, in whichever week they arrive, and the free first class doubles as the placement check that decides which batch that is.
Student work
Each of the four below is a live site you can open in a second tab right now. Students built them, students deployed them, and they were left online afterwards. Screenshots would have been easier to produce, which is exactly why we do not use them. More in Student Labs.
AI and ML
An assistant that takes a record of what someone has eaten and coaches them toward a target, carried from first idea to public URL by one student.
AI and ML
A tutoring chatbot that fields mathematics and programming questions, written and put into production by a student.
AI and ML
A tool that helps a young person work out when an online situation has turned unsafe, designed and shipped by a student.
Data and forecasting
A weather application running on live feeds, the closest relative in this set to the electricity price project above.
Fees
The United States is the one market we teach where nothing has to be converted. Our list prices are dollar prices, so what is printed below is what is charged: USD 40 and USD 100. Billing runs month by month. There is no enrolment charge and nothing here commits you to a year.
Free first class
$0
USD 0, no card needed
Group batch
$40
USD 40 a month
One to one
$100
USD 100 a month
One note on tax, since American buyers reasonably ask. There is no national sales tax on this price. Sales tax is set by states and localities, and whether live online tutoring is taxable differs between them, so the figures above are tuition and we print neither a tax-inclusive total nor a blanket claim that no tax applies. Either would be wrong somewhere.
What families say
Modern Age Coders holds a 4.9 average across 547 Google reviews. The six below are copied as the families wrote them. We do not draft testimonials and we do not pay for them.
★★★★★
"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
★★★★★
"My son has been attending this coding class for the past couple of months, and I've been genuinely impressed with both his progress and enthusiasm. The instructors are incredibly patient and knowledgeable."
Poonam Rathore
Parent
★★★★★
"I highly recommend this computer coding class! The teachers are incredibly knowledgeable and passionate about coding. They make every session engaging and insightful."
Ritu Kedia
Parent
★★★★★
"One of the most wonderful education centres out there. Education is not limited to school syllabus but focuses on skill development. Learning here has been a wonderful journey and still continuing."
Vansh Agarwal
Student
★★★★★
"Coding classes here make learning very interesting and conceptual. The teachers teach us in a very easy-to-understand and efficient manner."
Arush Poddar
Student
Questions from the United States
No. The United States has no national curriculum and no national exam board, because education is run by the fifty states and roughly 13,000 local school districts. The CSTA K-12 Computer Science Standards, published by the Computer Science Teachers Association, are the closest thing to a shared reference, and states adapt them. Code.org's 2025 State of AI and CS Education report, data finalised 1 December 2025, found only 4 of the 50 states explicitly emphasise artificial intelligence inside their computer science standards: Colorado, Virginia, North Dakota and Ohio. Executive Order 14277, signed 23 April 2025, directs federal agencies to build AI literacy for K-12 students. A federal instruction exists, 46 state standards are silent, and families close the distance themselves.
Neither is a machine learning course, which is worth knowing before choosing. AP Computer Science A is a programming course taught in Java, so it builds the coding discipline that later machine learning work sits on. AP Computer Science Principles is broader and more conceptual, and carries a through-course Create performance task submitted online rather than sat entirely as a written paper. Neither will have your child assemble a dataset, hold out test rows, train a model or report an error in real units. That is what our teenage track is for, and it runs beside whichever AP your school offers.
Two weekend slots reach both coasts and one weekday slot reaches the Pacific coast only. Saturday 11:00am Eastern, which is 8:00am Pacific, is the flagship batch. Sunday 12:00pm Eastern, which is 9:00am Pacific, is the second weekend cohort. Tuesday and Thursday at 6:30pm Pacific is the weekday option, Pacific only because 6:30pm Pacific is already 9:30pm Eastern. The part most providers will not say out loud: a weekday evening class at 6:00pm to 8:00pm Eastern falls between 3:30am and 5:30am in India, where our teaching operation sits, so we do not sell it. The United States observes daylight saving and India does not, so every American timetable is rebuilt on the second Sunday in March and the first Sunday in November.
Read what a strong programme publishes. Carnegie Mellon University's admission page for its Bachelor of Science in Artificial Intelligence asks for 4 years of English, 4 years of mathematics including at least algebra, geometry, trigonometry, analytic geometry and elementary functions and preferably calculus, 1 year of physics, 2 years of chemistry, biology or computer science, 2 years of a foreign language and 3 electives. That is a coursework profile, not a cut-off score, and the page publishes no minimum grade point average, no test threshold and no acceptance rate for the major. Mathematical depth is the binding constraint. Projects matter alongside it because an admissions reader can open a link and watch the mathematics being used.
Yes, and two are worth knowing about years before anyone applies. The University of Texas at Austin runs an online Master of Science in Artificial Intelligence of 30 credit hours, 3 hours of required courses plus 27 hours of electives across 10 courses, described by the university itself as an affordable, advanced degree priced at $10,000 plus fees, with international student fees and late registration fees possible. Fall applications open 15 December, priority 15 March, final 15 April; spring open 1 June, priority 1 August, final 1 September. Georgia Tech's Online Master of Science in Computer Science carries a machine learning specialisation, is also 30 credit hours, typically takes three years, requires no GRE, and prefers a related undergraduate degree with a cumulative grade point average of 3.0 or higher. Both screen on an undergraduate record.
The honest answer uses federal statistics rather than a vendor forecast. The Bureau of Labor Statistics Occupational Outlook Handbook puts the median annual wage for data scientists at $112,590 as of May 2024, with employment projected to grow 34 percent between 2024 and 2034, which the agency calls much faster than average, and about 23,400 openings projected each year across the decade. Computer and information research scientists sit at a median of $140,910 as of May 2024, with 20 percent projected growth and about 3,200 openings a year. Those are federal projections, and they describe the people who build these systems rather than the people who type at them.
Twelve states do, according to Code.org's 2025 report: Alabama, Arkansas, Indiana, Louisiana, Nebraska, Nevada, North Carolina, North Dakota, Rhode Island, South Carolina, Tennessee and West Virginia, the last being the most recent addition. No state requires both artificial intelligence and computer science for graduation, and only 5 of the 50 fund AI and computer science professional development for teachers: Arkansas, Indiana, Louisiana, New Jersey and Ohio. In the other 38 states computer science is an elective your child may or may not be offered, and anything past it is arranged outside the school day.
AP exams are the one genuinely national fixture, administered in schools across two weeks in May. In 2026 that was 4 to 8 May and 11 to 15 May, with AP Computer Science Principles on Thursday 14 May 2026 and AP Computer Science A on Friday 15 May 2026, both at 12pm local time, and the CSP Create performance task due 30 April 2026 at 11:59pm Eastern. College Board has published 30 April 2027 for the next Create deadline but had not released the full 2027 subject grid as of July 2026, so we treat May 2027 as two weeks in May with dates to be confirmed. Late April into mid May is a crunch and we lighten the load through it. Spring break is a single week in March or April and is not synchronised nationally, so we do not pause for it. June to August is our deepest window.
Start here
Leave a number and a mentor calls you back, usually within a few hours and at an hour that works in your time zone rather than ours. The first class is a real class with a real teacher, and doubles as a placement check so your child starts at their own level rather than the average one.
Prefer to look around first? Read how we teach, browse the full course catalogue, or see what students have built in Student Labs.