In 2024, about 1 in 6 developers in their first two years rated their job 4 out of 10 or worse. For developers with 21+ years it was about 1 in 10.
Basically, I worked as a junior developer at a mid-sized tech company for a few months before starting my Masters. The work didn’t really have an end time. There was always a task to finish, so 5pm didn’t mean done, and some nights I kept working on it at home. I was stressed most of the time and I’d get through the week just waiting for the weekend. I kept wondering if it was just me, or if every tech company works like this. Now I’m about to go back into tech, at a time when AI is changing what a junior developer does. So I wanted to find out who struggles the most in this industry, and if people starting out like me are the most at risk.
I used the Stack Overflow Developer Survey from 2024 and 2025. In 2024, 29,126 developers rated how satisfied they are in their current job from 0 to 10, and in 2025 it was 26,670. The survey doesn’t measure burnout directly. So I use low job satisfaction (a score of 0 to 4) as the warning sign, and in this report “unhappy” means that.
How to use the charts: hover over anything for exact numbers and sample sizes. Click legend items to switch groups on or off. Chart 1 has a year slider, chart 2 has a country dropdown, chart 4 highlights an experience group across both years when you click a bar, and chart 5 has year buttons.
Data: Stack Overflow Developer Survey 2024 and 2025 (Stack Overflow, 2024, 2025), downloaded via Kaggle (Alan, 2024; Aslam, 2025).
Text summary: in every country with 100+ respondents, between about 4% and 34% of developers rate their job 4 or less. Most countries sit somewhere between 5% and 15%. Grey countries had fewer than 100 respondents. The colour scale tops out at 25%, so hover for exact values.
Across all developers, 12.5% were unhappy in 2024 and 9.2% in 2025. The slider shows the map shifting a bit between years, but there is no country where nobody is unhappy. India, where I’m from, went from 13.6% to 9.9%. Australia, where I live now, went from 12.3% to 6.8%. Location explains some of it, but most countries land in a similar range. So the better question is who is unhappy, not where.
Narrative technique: this chart is the hook. It sets up the question that the rest of the report answers (structure). The year slider is the interactivity, and both years use the same colour scale so the change between them is honest.
Design choice: I mapped the share of unhappy developers and not the raw count. Raw counts would make the US and India look worst just because they have the most respondents. Both years use one fixed colour scale so the slider shows real change, and countries with under 100 respondents are greyed out rather than shown with an unreliable number.
Data: Stack Overflow Developer Survey 2024 and 2025 (Stack Overflow, 2024, 2025), downloaded via Kaggle (Alan, 2024; Aslam, 2025).
Text summary: in all three work setups, the share of unhappy developers goes down as experience goes up, in both years. The highest point is in-person developers with 0 to 2 years of experience in 2024, at about 20%. Remote and hybrid juniors sit lower.
The pattern is the same in both years and in every work setup. The first few years are the hardest and it slowly gets better with experience. Splitting by work setup shows which juniors have it worst. In-person juniors were the highest, about 1 in 5 in 2024. The dropdown switches to the United States or India, where I’m from. The error bars get wider there because the groups are smaller, and any group with fewer than 30 people is hidden so it isn’t shown with an unreliable number.
One thing to note: 2025 is lower overall, but the 2025 sample also has more experienced people in it. That’s why I compare experience groups with each other and not only the two yearly totals.
Narrative technique: the grey band marks the junior years in every panel and the callout points to the worst group (highlighting and annotation). The country dropdown is interactivity with a purpose, because it lets the reader check if the pattern holds in their own country. Showing both years together rules out the idea that it was just one bad year.
Design choice: I used small multiples, one panel per work setup with the same y-axis, instead of six lines in one chart. This keeps each panel readable and lets the panels be compared by position. Experience is ordered so I used lines, and the y-axis starts at 0 so the drop isn’t exaggerated. The error bars are 95% confidence intervals, so it’s clear which differences are solid and which come from small samples. The years are grey dashed and black solid on purpose, so orange only means “unhappy” or “worse-off” across the whole report.
Data: Stack Overflow Developer Survey 2024 and 2025 (Stack Overflow, 2024, 2025), downloaded via Kaggle (Alan, 2024; Aslam, 2025).
Text summary: in both years, the darkest cells are in the bottom row (bottom third of pay for their country), mostly in the early-career columns. Moving up to the top third of pay, or gaining experience, both go with lower shares of unhappy developers.
I ranked everyone’s pay inside their own country, because a low salary in US dollars mostly just means someone lives in a cheaper country. Even with that, the bottom third of pay is the unhappiest row at almost every stage. And most juniors are in that row: 60% of juniors in 2024 and 66% in 2025 were in the bottom third of pay for their country. So for people starting out, low pay and being early in their career add up.
Narrative technique: the bottom pay row is outlined and has the darkest cells, so the reader’s eye goes there first (highlighting). Each cell is labelled with its number, so the message doesn’t depend on reading colour.
Design choice: the heatmap shows pay, experience and year together, and both years share one colour scale so the panels compare fairly. Ranking pay inside each country stops currency and cost of living from creating a false pattern. Every cell is labelled so the chart still works for colour-blind readers.
Data: Stack Overflow Developer Survey 2024 and 2025 (Stack Overflow, 2024, 2025), downloaded via Kaggle (Alan, 2024; Aslam, 2025).
Text summary: at every experience level and in both years, developers who say AI is a threat to their job are the most likely to be unhappy, usually close to double the “No” group. The gap is biggest for the most junior developers in 2024.
Honestly, I didn’t expect this one. Developers who see AI as a threat to their own job are the unhappiest group at every experience level. That group also grew, from 2,380 people (10.7%) in 2024 to 3,877 (14.6%) in 2025. This doesn’t prove that worrying about AI causes unhappiness. It could be the other way around, and unhappy people may just worry more. But a lot of entry-level work is the kind of task AI tools now do, so for juniors it adds to an already hard start.
Narrative technique: orange for “Yes” against blue for “No”, with “Not sure” in grey so it fades back (highlighting). The “x times” labels state the size of the gap directly (annotation). Clicking a bar uses linked highlighting, so the same experience group lights up in both years and is easy to compare. This chart is the turning point of the story (transition), moving from the older problems (experience and pay) to the new one (AI).
Design choice: the bars start at 0 because bar length is the encoding. The two years share a y-axis so the heights compare directly. The colours come from the Okabe-Ito colour-blind safe palette, with “Not sure” in grey so the Yes and No comparison stands out. The thin lines are 95% confidence intervals. For “Yes” and “No” they don’t overlap at any experience level, so the gap isn’t down to chance.
Data: Stack Overflow Developer Survey 2024 and 2025 (Stack Overflow, 2024, 2025), downloaded via Kaggle (Alan, 2024; Aslam, 2025).
Text summary: among junior developers, the biggest gap in both years is between those who see AI as a threat to their job and those who don’t. Pay (bottom vs top third in their country) and work setup (in-person vs remote) also show clear gaps, and the worse-off side is always above the junior average.
This chart brings the story together. It only includes juniors, the group that came out worst in chart 2. Each line is the gap between the worse-off and better-off juniors on one factor. AI worry has the biggest gap in both years. Pay and work setup come next, and they swap places between 2024 and 2025. The buttons switch between years.
So the answer to my question is yes. People starting out in tech right now are the most likely to be unhappy. But it’s not only about being junior. Low pay, less flexibility and worrying about AI all go along with it. These charts don’t prove cause and effect. But if I was picking my next job in tech, these are the three things I would look at.
Narrative technique: this is the ending, so it brings the earlier charts together into one takeaway (messaging). The title and the “All juniors” reference line tell the reader what to compare.
Design choice: I used a dumbbell plot so each pair sits on the same position scale and the gaps are easy to compare. The worse-off group is an orange diamond and the better-off group is a blue circle, so both shape and colour carry the meaning. The dotted line marks the average junior as a reference.
survey_results_public.csv from each year directly, and no
data is typed in by hand.Alan, B. (2024). Stack Overflow annual developer survey 2024 [Data set]. Kaggle. https://www.kaggle.com/datasets/berkayalan/stack-overflow-annual-developer-survey-2024
Anthropic. (2026). Claude (Opus 5.5) [Large language model]. https://claude.ai
Aslam, A. (2025). Stack Overflow developer survey 2025 [Data set]. Kaggle. https://www.kaggle.com/datasets/aliaslam25/stack-overflow-developer-survey-2025
Plotly Technologies Inc. (n.d.). Plotly R open source graphing library. https://plotly.com/r/
Stack Overflow. (2024). Stack Overflow developer survey 2024 [Data set]. https://survey.stackoverflow.co/2024/
Stack Overflow. (2025). Stack Overflow developer survey 2025 [Data set]. https://survey.stackoverflow.co/2025/
I used Claude (Anthropic, 2026) to help troubleshoot R errors, clarify doubts about the code, and improve grammar and wording. I selected the topic and made the decisions about the story and its direction. The analysis, chart creation, and interpretation of the results were completed by me, and I checked the final output myself.