Talks and events

What I speak about

I speak to a few hundred students a term and none of it was visible anywhere. This is the record, and it is also what an organiser needs before inviting me.

Sessions

Two of these I have given, on the dates shown. The other three are drafted from what I research and teach, not from a session I have delivered — they are a starting point for a conversation rather than a fixed offer, and the formats are mine to confirm.

Your data, your privacy: understanding digital footprints

What a phone, a search box and a loyalty card actually record, why removing your name from a dataset does not make it anonymous, and what a student can do about it. No mathematics assumed.

Given before
Undergraduates and school students, any discipline
60 minutes, with questions
Delivered to 800 students in January 2026

RAG, vector databases and semantic search

How retrieval-augmented generation works end to end: embeddings, a vector store, the retrieval step, and where the failures actually come from. Built for faculty who need to teach it next term.

Given before
Faculty development programmes, postgraduate students
Half day, hands-on
Delivered at a faculty development programme, GCT Coimbatore, May 2025

Differential privacy, from the Netflix Prize to the census

The de-anonymisation results that broke the old approach, the guarantee that replaced it, and what it cost when the United States Census adopted it. The lecture version of my doctoral work.

Proposed, not yet given
Computer science students and researchers
45 to 60 minutes

Privacy by design in a machine learning pipeline

Where privacy is won and lost in a real system: collection, training, model release, deployment. Aimed at people who are building something now and have to make these decisions this month.

Proposed, not yet given
Engineering faculty, industry teams, project students
45 to 90 minutes, or a workshop

Responsible AI for engineering students

Fairness, interpretability and accountability as engineering constraints rather than ethics slides — what to measure, and what to do when the measurement comes back badly.

Proposed, not yet given
Undergraduate and postgraduate students
45 minutes

On the record

5 sessions and events. Across the 2 where a headcount was taken, that is 920 people in the room.

31 January 2026
Your Data, Your Privacy: Understanding Digital Footprints and Privacy (guest lecture)Coordinator · 800 in the audience
31 January 2026
Official Pre-Summit Event, AI Impact Summit 2026 — AI for Indian Knowledge SystemsHost
24 October 2025
ACM Student Chapter Inaugural FunctionCoordinator · 120 in the audience
21 May 2025
Guest lecture on RAG, vector databases and semantic search, FDTP at GCT CoimbatoreResource person
2025
ACM India Chapter Summit 2025, NMIMS IndoreDelegate

Roles I hold

Standing commitments rather than one-off events. These are where most of the sessions above come from.

Since Oct 2025
PSG iTech ACM Student ChapterFaculty Coordinator
Since Jul 2026
ACM COMPUTE 2026 Working Group: Understanding Refute Problem Difficulty for Learners in Indian Computing EducationCo-lead
2 reviews
Engineering Applications of Artificial Intelligence (Elsevier)Reviewer

Inviting me

Tell me the audience, the date and how long you have, and I will tell you within a few days whether I can do it. The work with me page has the form, and the speaking path there lists what I need to know.

For anything outside Coimbatore, say in the first message what you can cover on travel and what the date is fixed to. It saves an exchange of messages later.

Still to come: slides and photographs from the sessions above. I have them, they are not yet gathered in one place.