Placement Cell Automation: Realistic Scope

The training-and-placement officer had two thousand résumés, eighty companies, and a spreadsheet that crashed on Fridays. A vendor offered to 'match talent to roles with AI'. The matching features, when we asked, included CGPA, a scraped GitHub, and a 'communication score' from a one-minute video. They did not include the fact that several recruiters had asked, off the record, for a 'certain background'. The TPO did not need a smarter prejudice. He needed slots, eligibility lists that matched the published notice, and a way to stop emailing the wrong branch.

This guide is the realistic scope for Indian campus placement cells — public universities, NITs, state technical campuses, and private colleges that still run a central TPO. It is not a product tour of HR-tech. Those products are built for companies. You are a fiduciary for students, often a public authority, and sometimes a reservation-bearing institution even when the company is not.

Not legal advice. Company-side labour law and your placement policy both apply. Write the policy before the agent.

The clerk, not the recruiter

Safe verbs: publish the job notice as the company sent it; check student-uploaded fields against the published eligibility; book slots; answer 'when is the test' from the notice; remind students who opted in; produce the attendance sheet the TPO already signs.

Unsafe verbs: rank students for a company; predict 'likelihood of joining'; score spoken English from a video; hide a student from a recruiter because the model thinks the CGPA is stale; rewrite a résumé into a story the student did not write; send a pack to a company that is not in this drive.

The published notice is the law of that drive. If the company said 6.5 CGPA and no active backlogs, the agent flags against that. If the company said 'from any background', the agent does not add a hidden IIT-filter because last year those students converted. Last year is not this notice.

If the company did not write it, the agent does not add it.

Task	Do	Do not
Eligibility	Boolean against the published notice	Soft rank, 'culture', face, accent, GitHub stars
Consent to share	Per drive, withdrawable, logged	A one-time 'I agree to placement' at admission for all future companies
Résumé pack	Send what the student uploaded, to the named company	Enrich with social media or a library circulation profile
Interview practice	Optional, disclosed, not scored into the official list	A mandatory 'AI interview' whose score goes to the company
Diversity / reservation	Follow the campus policy and the law that applies to that employer	Let a model silently 'balance' or silently 'unbalance' a list

Consent is per drive, not a tattoo from first year

Sharing a résumé, category certificate, photograph and phone number with a named company is a new purpose. Admission is not that purpose. A placement-registration consent can be broader than one company if it is specific enough and withdrawable. 'Share with any recruiter forever' is not specific.

Category certificates in a placement pack are a flashpoint. Some public employers need them. Many private ones do not, or should not. The student decides what goes in the pack for that drive, within the notice. The agent does not attach the admission-time caste field because it is sitting in the ERP.

Minors in diploma and integrated programmes need the Section 9 conversation again. Do not invent a 'campus exception'. Ask counsel and read the live Rules.

Companies are not your processor — and vendors are not companies

When you send a pack to a recruiter, you are sharing with another fiduciary for their hiring. Write that. When you send a pack to a 'placement cloud' that trains on résumés, you have added a processor who wants to become a product. Refuse the training. Prefer on-prem. If a company insists on their ATS, send only the opted-in pack for that drive.

Do not let the agent auto-create student accounts on random HR platforms. That is how personal data goes to a start-up that dies in eighteen months and sells its disk.

Off-campus 'opportunities' scraped from the web are a moderation problem. Fake internships and data-harvesting forms are ordinary. A human still approves a notice before the agent publishes it.

Metrics that do not lie

Measure time-to-publish a notice, eligibility-check errors caught, and no-show rates. Do not measure 'AI match accuracy' unless you are prepared to define match without discrimination. Placement percentage remains a human statistic with a denominator the IQAC will fight about. Do not let the agent pick the denominator.

Internships and PPOs are not the same drive

A summer internship notice and a final-placement notice are different purposes even when the company is the same. Consent, packs and eligibility must be separate. A student who shared a résumé for a two-month internship did not agree to a pre-placement offer campaign six months later. The agent should not 'helpfully' reuse last season's pack.

PPOs that arrive by a recruiter's personal mail still need the TPO as publisher if they are to use the official list. Otherwise you will have a shadow process the reservation cell and the IQAC cannot see. The agent has no business ingesting a recruiter's private spreadsheet of 'people we liked'.

International offers add a transfer and a foreign-employer notice. Do not let the agent create accounts on unknown overseas ATS platforms. If the student opts in, send the pack they saw. If they do not, the company writes to them directly without your file.

Objections you will hear — and what to do with them

These are the lines that stall the file. Answer them in the room, then put the answer in the note. A spoken answer without paper will be forgotten by the next officer.

Companies demand a ranked list by tomorrow.

Give them the published eligibility list in the order the notice described — often CGPA or registration order — and say so. A secret model rank is still your act. If they walk, they walked because you would not invent a criterion.

We need AI to be fairer than humans.

Then publish the rule. Fairness that cannot be written into the notice is a vibe. Historical placement data will encode last decade's prejudice if you train on it.

Students want a ChatGPT résumé rewrite.

Offer optional language help with a disclosure, on-prem, no training, student accepts the text. Do not submit a rewritten résumé the student has not seen. Do not score the rewrite.

A CRM is the real need.

Then say CRM, budget it, and staff it. An agent is not a free Salesforce. Alumni and companies still need a human owner.

One placement season without a ranker

Pick the next drive, not the whole season, if you are late.

1. Week 1: rewrite the placement policy — verbs, per-drive consent, no ranker, no video score unless advertised.
2. Week 2: connect the agent to notices, eligibility fields, and the slot calendar only.
3. Week 3: student notice and withdrawal path. Category fields off unless the drive needs them.
4. Week 4: run one drive. Audit every pack that left campus. Expand only if the audit is boring.

How this shows up in the file

Placement policy addendum: the agent is a clerk. Companies receive opted-in packs. No training. No partners toggle. No model rank. TPO remains the publisher of every notice.

Attach the DPDP basis and the retention for rejected-drive résumés. 'We might need them next year' is not a retention schedule.

This article is informational field guidance for Indian public universities and colleges, not legal, regulatory, academic, labour or procurement advice. Confirm against the live AICTE / UGC / NAAC notice, the DPDP Act and Rules as commenced, your Act and statutes, reservation orders, labour counsel and the syndicate before you file it.

How this works on an Indian campus

A P3 University should be able to run “Placement Cell Automation: Realistic Scope” without importing a US playbook. “placement cell automation” hits UGC/AICTE/NAAC clocks, exam secrecy, reservation rules, and students who may be minors.

A placement agent is a clerk for calendars and eligibility. It is not a recruiter, not a caste-blind ranker, and not a CRM you never staffed. DPDP applies to student personal data. Chatbots are not a strategy. Exam and admissions writes stay with officers. Affiliated colleges need isolation, not one shared index.

• No production student data in a vendor SaaS sandbox.
• Write the academic integrity policy before the tool.
• Consent and purpose tags on student-facing agents.
• Budget for staff training, not only licences.

Close this loop before the next CAB

Put “Placement Cell Automation: Realistic Scope” on the next change-advisory or bid-opening agenda as a single line item with an owner. If it cannot earn a line item, it will not earn a control. The owner should be a P3 University, not “the vendor.”

Revisit the item when the model, the GeM term, the region, or the SI changes. “placement cell automation” is not a one-time workshop. It is a watch item. Date the last check. Unsigned watch items are souvenirs.