Editorial, Corrections and Fact-Checking Policy

On this page

• 1. Editorial Policy
• 1.1 Who writes this, and under what name
• 1.2 How we use AI in research and writing
• 1.3 What we cover, and what we don't
• 1.4 House rules that apply to every article
• 2. Fact-Checking Policy
• 2.1 What counts as a source
• 2.2 Measured facts vs. projections
• 2.3 What we do when we cannot verify something
• 3. Corrections Policy
• 3.1 How to report an error
• 3.2 How we fix it, and how dates change
• 3.3 What a correction is not

1. Editorial Policy

1.1 Who writes this, and under what name

Articles on this blog are published under the byline of Aniruddha Abhay Kalbande, founder of Fireblaze AI School. His background is IT consulting and technical training; company-specific interview-preparation guides and career and learning guides sit within that expertise. We are conscious that deep, specialist analysis of unrelated industries carries a single byline further than one person's own direct expertise reasonably extends, and we address exactly that in the AI-use section below rather than pretend it does not apply.

1.2 How we use AI in research and writing

Yes, AI tools are used in producing this blog, and we would rather state that outright than have a reader discover it and wonder what else we did not mention. In practice, this looks like: an AI tool researches published sources on a topic (a company's interview process, a labour-market statistic, an industry development), drafts an article from that research, and a human reviews it before it goes live. This is different from, and less trustworthy than, first-hand reporting, and we do not dress it up as personal experience it is not. Where an article describes an interview process, it is describing patterns reported by candidates and recruiters in public sources, not a claim that our founder or staff sat in that interview.

What this means in practice:

• Claims of fact are expected to trace to a citable source. An AI-assisted first draft does not lower that bar; if anything it raises the need to check it, since a fluent sentence is not the same as a true one.
• We do not use AI to generate fabricated statistics, testimonials, salaries, placement outcomes or client names, regardless of how plausible the output looks. This is a standing rule for every article on this site, not a case-by-case judgement call.
• Large batches of new or rewritten articles are not published simultaneously with a shared publish date. A page's datePublished reflects the date that version of the content actually went live, and does not get reset by an unrelated site-wide change such as a template or navigation update.
• We would rather publish fewer, better-checked company interview guides than many thin ones. Where we cannot find anything genuinely specific to say about a company's process, we say so, or we do not publish a standalone page for it.

1.3 What we cover, and what we don't

Our blog focuses on career roadmaps, choosing a course, interview preparation by company and by topic (SQL, Power BI, statistics, machine learning), salary and placement outcomes, and how Generative AI is changing entry-level data and development work. We also publish broader AI industry analysis, including sector coverage (such as aerospace and defence AI) that sits further from our core teaching expertise. Where that is the case, we lean more heavily on the sourcing and gap-disclosure standards in section 2, precisely because the byline's authority is thinner there.

1.4 House rules that apply to every article

• No fabricated numbers, ever. Placement rates, salaries, learner counts, testimonials and company names are either verified and cited, or gated and marked as unverified. They are never invented to fill a gap.
• No em dashes in body copy. A small, deliberate style choice, consistent across the site.
• Ground-truth facts stay identical everywhere. Founded 31 October 2017, NASSCOM partner for the PGP in Data Science & Analytics only, three Nagpur centers, fees include 18% GST. These do not vary between pages.
• One real article per publish, dated the day it actually went live. We do not backdate content to look older, and we do not front-load a batch of drafts with the same date to imply a publishing cadence that did not happen.

2. Fact-Checking Policy

2.1 What counts as a source

In order of preference:

1. Primary sources. Government data and registers, a company's own filings, statements or press releases, court judgments, official policy documents.
2. Named studies with a stated methodology and sample. NASSCOM reports, LinkedIn's Jobs on the Rise and Skills on the Rise, Naukri's JobSpeak index, Indeed Hiring Lab, PayScale and Indeed salary data, the Stanford HAI AI Index, and comparable named, dated research.
3. Established publications reporting on the above, used when a primary source is paywalled or otherwise inaccessible, and always attributed to the outlet that reported it, not presented as if we read the primary document ourselves when we did not.

We do not treat a number as sourced just because it appears on many websites. Several figures that circulate widely about the Indian AI job market, including a specific claim about the size of the AI skills gap, could not be traced back to a primary document stating that figure, and we do not repeat them. The AI Skill Index of India publishes its own list of exactly which widely-repeated claims we checked and rejected, alongside why.

2.2 Measured facts vs. projections

We distinguish, explicitly, between something that was measured (a survey result, a job-posting count, a reported salary) and something that is projected or forecast (an estimate of where a market will be in a future year). Both can be useful, but they are not the same kind of claim, and an article should never let a forecast read as if it already happened. Where we quote a projection, we say who made it and for what date.

2.3 What we do when we cannot verify something

We publish the gap, not a guess. If a reasonable, well-sourced figure exists for a related question but not the exact one asked, such as a role's salary range in a city where no dataset exists, we say that plainly rather than interpolating a number that looks reasonable. Two small samples do not make a reliable average, and we would rather tell you a data point had 11 responses behind it than round it up to sound authoritative.

3. Corrections Policy

3.1 How to report an error

Email corrections@fireblazeaischool.in with the article's URL and a description of what you believe is wrong, ideally with a source we can check. We read every message and reply, including when we disagree that something is an error.

3.2 How we fix it, and how dates change

A confirmed factual error is corrected within 5 working days of us being able to verify the report. When a specific article is substantively corrected, that article's dateModified is updated to the real date of that fix. It is not bumped for unrelated reasons, such as a sitewide template, navigation or styling change, and we do not mass-update dates across many articles at once to manufacture an appearance of freshness. A correction to fact is different from a routine update (a batch start date, a current fee, a scholarship deadline); routine updates are made without a special notice, since they describe things that are expected to change on a schedule, not something we got wrong.

3.3 What a correction is not

A correction is not a mechanism for removing a true but unflattering fact, and it is not a way to backdate content to claim it is older than it is. If new information genuinely changes a conclusion, we would rather update the article and say so than quietly delete the old claim.