AI Project on Resume or Portfolio? The 2026 Hiring Tradeoff

Recruiters in 2026 Indian hiring funnels see your resume first, your portfolio second, and your demo link only after they have already decided you are worth a deeper look.

That sequence matters more than the content of any single artifact. The same AI project can be represented as a two-line resume bullet, a detailed portfolio page, or a live interactive demo. Which one you lead with determines whether a recruiter keeps reading or closes the tab. This article explains how to use all three correctly.

Resume and portfolio serve different funnel stages

The resume is a sequence of filters, not a document anyone reads start to finish. An applicant tracking system parses the uploaded text, matches it against role keywords, and assigns a score. A recruiter who shortlists from that pool then scans each passing resume for 30 to 60 seconds before deciding to pass or reject. At no point in that sequence does a recruiter follow the GitHub link embedded in bullet point three.

The portfolio is a different artifact entirely. It is what the recruiter, hiring manager, or technical reviewer opens after they have already decided you are worth a second look. Unlimited space. Rich narrative. Architecture diagrams, code snippets, outcome metrics with full context. The portfolio holds everything the resume cannot.

These are not competing surfaces. They serve sequential stages of the same funnel. Treating them as either-or is the mistake most freshers make: they either cram the project into a long resume bullet that no recruiter reads fully, or they skip the resume entry and hope the recruiter finds the portfolio on their own. Neither works.

What belongs on the resume (and what to cut)

One AI project on the resume means at most two bullet points plus one link. The bullet pattern that works:

• What the system does in plain terms, not jargon
• The tech stack, concise rather than exhaustive
• A measurable outcome — accuracy, latency, error rate, test coverage, user count

The difference between a bullet that passes recruiter review and one that does not:

Version	Text
Weak	”Built a chatbot using Python and LangChain”
Strong	”Built a customer intent classifier using LangChain and a fine-tuned sentence-transformer; 89% accuracy on a 5,000-query test set; deployed on Streamlit Community Cloud”
Weak	”Worked on an NLP project using HuggingFace”
Strong	”Fine-tuned a BERT-base model on 12,000 labelled product reviews for sentiment analysis; F1 of 0.84; evaluation notebook and inference code on GitHub”

The strong versions share three properties: they name the specific approach, they state a number, and they give the reader something to follow up on.

The link question

GitHub repository or portfolio page? Both if the resume has space. One if it is already tight. When forced to choose, prefer whichever has the cleaner entry point: a readable README beats a portfolio page that is still half-finished, and a polished project page beats a repo with a single commit labelled “first commit.”

What ATS systems do with links: Most applicant tracking systems parse the plain text of your uploaded PDF and do not follow embedded hyperlinks. The link is invisible to ATS scoring; it is there for the human recruiter who opens your file after shortlisting you. Write the bullet to pass ATS. Include the link for the recruiter.

What to cut: a Skills section that lists “Python, Pandas, NumPy, scikit-learn” without any context. Move those names into the project bullet itself, where the context makes them meaningful. A standalone tools list at the bottom of the resume adds almost nothing on its own.

When the portfolio takes over the conversation

After shortlisting, the funnel changes shape. The hiring manager or technical reviewer wants depth, and that is when the portfolio link gets clicked.

How this plays out differs by platform:

• On LinkedIn, the portfolio link lives in the Featured section or inside the Experience entry for the project. Hiring managers visiting your profile typically look there after reading the headline and current experience.
• On Naukri, the “Online Presence” section holds portfolio and GitHub URLs. The recruiter flow is: resume download, profile scan, online presence links. The portfolio is one click away from the ATS-parsed resume, not buried several pages deep.
• On Wellfound (formerly AngelList Talent), the application form has structured fields for GitHub URL, portfolio, and personal site. Startup hiring managers on Wellfound expect these fields to be filled in. A blank GitHub field on a Wellfound application creates friction that it would not on other platforms.

These are not judgments about which platform serves freshers better. They are structural observations about where the portfolio link surfaces in each hiring flow. Meeting recruiters where they are already looking costs nothing.

A strong AI project portfolio page has five elements:

1. Problem — why does this exist, and what was the gap it fills?
2. Approach — what decisions did you make, and why those choices over obvious alternatives?
3. Results — accuracy, latency, user test feedback, any measurable outcome
4. What you would change — the section most student portfolios skip entirely
5. Links — GitHub repo, live demo, dataset source if public

The “what you would change” section is the differentiator. A fresher who can articulate the limits of their own work is more credible to a hiring manager than one who presents only wins. It signals that you understand engineering involves tradeoffs, not just implementations.

When the interview arrives, the question “walk me through your AI project” is essentially asking you to narrate these five sections out loud. FACE Prep’s structured answer guide for that interview question covers the 5-part verbal pattern that maps directly to the same sections.

The demo link: your third artifact

A live demo answers the one question that a resume bullet and a portfolio page cannot: did this actually run?

For AI projects, the demo matters most when the output is something a non-technical recruiter or hiring manager can experience without reading code. A sentiment analyser that accepts pasted text. An image classifier with drag-and-drop input. A simple chatbot with a working UI. These demos convert scepticism into belief faster than any benchmark number in a PDF.

Hosting options that freshers use in 2026:

• Hugging Face Spaces — free tier supports Gradio and Streamlit apps; widely recognised by ML-aware hiring teams in India and globally
• Streamlit Community Cloud — free for public apps; straightforward deployment if your project is already a Streamlit app
• Vercel or Netlify — for Next.js or React frontends built over a model API

The honest risk: demos break. APIs get deprecated. Free-tier usage limits get hit. If you lead with a demo link in an application and it is down when the recruiter clicks it, you have created a worse impression than no link at all.

The practical fix: keep a 2-minute screen recording of the demo working correctly. If the live link fails during an interview, the recording is your fallback. If you mention a demo URL in your resume, check that it still works on the morning of each interview.

One more calibration: not every AI project needs a live demo to be worth applying with. A well-documented research-style project (fine-tuned model with a thorough evaluation) can carry its weight through the GitHub README and portfolio page alone. The demo is valuable when it adds something the static pages cannot show, typically interactivity and direct evidence of deployment.

What makes an AI project worth linking to

The gap between a project you built to understand a concept and one you would confidently link to in a job application comes down to three things: a clean README, a measurable outcome, and some evidence it ran.

A Jupyter notebook with no output cells, a GitHub repo with a single commit labelled “initial”, or a Streamlit app that requires the recruiter to install dependencies locally: none of these are ready to link to. A deployed app or a well-evaluated model with clear documentation, a stated result, and a live or recorded demo is.

Two complete, link-ready projects tell a hiring manager more about your engineering ability than a list of ten completed online courses. Not the tools used in isolation. Not the platforms that issued the certificates.

FACE Prep’s 2026 AI roadmap for Indian engineering students maps the skill sequence from fundamentals to a first deployed project. Start there. For interview prep, common anti-patterns in first AI engineering interviews covers what typically goes wrong when freshers overstate a project’s scope.

The argument this entire article builds to: one deployed AI project with a clean README and a working demo URL tells a hiring team more than a stack of completion certificates. TinkerLLM is where that first project gets built. At ₹299, it puts real LLM API calls in your hands, and the micro-project you ship is what goes into the resume bullet and portfolio page described above.