Jayin Khanna
A HEAD FULL OF DREAMS
Mathematics | Deep Learning | Philosophy | Quant Finance
About Me
Jayin Khanna
Education
- BSc (Research) in Mathematics — SNIoE, May 2026
Minor in CSE · Specialisations: AI & ML · Mathematical Finance - BS in Data Science & Applications — IIT Madras (ongoing)
Honours & Credits
- Program Batch Topper — SNIoE
- 2nd Prize, Best Thesis — SNIoE
- 238 credits across 4 years (176 SNIoE + 52 IIT Madras) — view course list
My Research Interests
- Core & applications of Generative Models (DDPMs, Flow Matching); T2I and T2V models
- AI Interpretability and Alignment
- Representation learning
- Unsupervised and self-supervised learning
Exploring
- Theory of Deep Learning (slow steady reading)
- Mechanistic Interpretability
- Geometric Deep Learning
I work on interpretability, alignment, and safety in generative models and LLMs — with a mathematical bias toward understanding why methods work before scaling them. My current focus is safety alignment in text-to-image/video diffusion and trustworthiness repair in fine-tuned LLMs.
I'm looking for a long-term research project aimed at publication at ICLR, ICML, NeurIPS, or TMLR.
Current Research
Safety-guided flow matching for T2I/T2V diffusion — AI Institute of South Carolina (Dr. Amit Sheth, Dr. Amitava Das). Developing a safety-potential-guided rectified flow-matching formulation in CLIP embedding space to reduce harmful generations, benchmarked on the DETONATE dataset against TRCE, CURE, SAEUron, and DoCo.
Post-hoc trustworthiness repair in LLMs — Deep Representation Learning Lab, IISc (Dr. Prathosh A. P.). Combining EK-FAC curvature estimation with targeted gradient ascent to mitigate bias, unethical outputs, and toxicity in fine-tuned models — a compute-efficient alternative to retraining or RLHF, evaluated across Qwen2 and Pythia with submodular subset selection for non-redundant repair signals.
My undergraduate thesis on generative models for unsupervised speech time-scale modification, under the joint supervision of Prof. Prasanta Kumar Ghosh (SPIRE Lab, IISc) and Prof. Niteesh Sahni (SNIoE), won 2nd Prize for Best UG Thesis. I document this work as detailed technical notes rather than leaving it in notebooks, posted on my Research page (DDPMs, VAEs, GANs, contrastive learning).
My prior research spans:
- Generative models for Unsupervised TSM
- Depth-Conditioned Video Generation using ControlNet & AnimateDiff
- Attribution/interpretability methods (IIT Kharagpur, Prof. Niloy Ganguly)
- Developing ML models for sEMG at DRDO-INMAS
- Statistical time-series modeling (UC Santa Cruz ISRP, Prof. Bruno Sansó)
- Summer research programs: MTTS 2024 and Polymath Jr. 2024, 2025
I love doing research! Check out My Journey, Academic Development & Projects.
Beyond academics, I love playing football and running. I also love reading — mainly psychology and philosophy, but I explore other genres as well.
If you have cool project ideas, or want to discuss thought experiments or ideologies — regardless of the domain — and want to collaborate, ping me! I am always looking forward to interesting stuff!
My Experience
Research Assistant
Deep Representation Learning Lab, Indian Institute of Science (IISc), Bangalore
Supervisor:Dr. Prathosh A. P.
Period: Jun 2026 – Present
Working on enhancing trustworthiness in fine-tuned LLMs by implementing the EK-FAC preconditioned gradient ascent pipeline for post-hoc bias repair in Qwen2 and Pythia LLMs using Dolly for curvature estimation and CrowS-Pairs for targeted repair gradients.
Research Intern
IRT Group, AI Institute of South Carolina | DETONATE Group
Supervisor:Dr. Amit Sheth & Dr. Amitava Das
Period: Feb 2026 – Present
Working on safety alignment of Text-to-Image & T2V diffusion and flow matching models using a neurosymbolic approach called scene graphs.
Machine Learning Research Fellow
SPIRE Lab, Indian Institute of Science (IISc) Bangalore
Supervisor:Prof. Prasanta Kumar Ghosh
Period: Jul 2025 – Present (Summer Research Fellowship 2025)
Research on unsupervised speech TSM using Generative Models. Working with VAEs, GANs, Diffusion, and Flow Matching based speech TSM models to preserve speaker identity, naturalness and intelligibility.
Deep Learning Research Intern
Indian Institute of Technology (IIT) Kharagpur
Supervisor:Dr. Niloy Ganguly
Period: 2025
Worked on developing attribution techniques using Integrated gradients, Manifold IG, Guided IG towards neural network interpretability.
Machine Learning Research Intern
Institute of Nuclear Medicine and Allied Sciences-DRDO, Ministry of Defence
Supervisor:Dr. Shilpi Modi
Period: 2024
Developed a Convolutional Neural Network (CNN) ResNet model for classifying sEMG stress measurements, focusing on improving accuracy and generalization.
Machine Learning Research Intern
Institute of Nuclear Medicine and Allied Sciences-DRDO, Ministry of Defence
Supervisor:Dr. Shilpi Modi
Period: Present
Working on the application of network control theory to understand cognitive state transitions in the brain.
Statistics Research Intern
University of California Santa Cruz, CA (ISRP)
Supervisor:Prof. Bruno Sansó
Period: 2024
Conducted advanced research in Analysis of Time-Varying Quantiles for Environmental Variables, under the mentorship of Professor Dr. Bruno Sansó.
Machine Learning Intern
RightProfile by Syntellect
Period: Dec 2024 – Present
Part of the Research and Development team to develop Computer Vision and Deep Learning models to automate the annotation of 10,000+ raw images.
Machine Learning Intern
The Habitats Trust
Period: Dec 2024 – Present
Conducting research and development on modern Computer Vision and object detection models such as MegaDetector, Zamba, and Timelapse to classify and analyze wildlife in camera trap images.
Academic Development
Reviewer, NeurIPS 2026
Conference on Neural Information Processing Systems
Polymath Jr, Summer Research Intern
June – Aug 2024, 2025- Engaged in research on Generative AI and optimal transport, Non-Local Models across two years.
- Focused on developing and analyzing mathematical models that incorporate non-local interactions & Optimal Transport based generative models.
Mathematics Training and Talent Search Program (MTTS 24)
May 2024- Selected from over 3,000 applicants all over India for one of 180 seats. Completed rigorous coursework and led discussions in Linear Algebra, Real Analysis, Proof writing and Number Theory.
- Engaged in daily discussions with leading professors and mentored peers during problem-solving sessions.
Talk Series (3 Talks): "Generative AI: A Mathematical Exploration of How Machines Learn to Create"
- Delivered a first-principles, mathematical explanation of VAEs, GANs, and Diffusion Models.
Teaching Assistant for MAT161: Applied Linear Algebra
Jan – Mar 2025Student Tutor for MAT101: Calculus I
Aug – Dec 2025Undergraduate Thesis Expo
Apr 2026- Presented undergraduate thesis to a jury and the Departments of Mathematics and Computer Science; awarded 2nd Prize for Best UG Thesis.
My Skills & Expertise
Programming Languages
By this time, It'd probably be a crime to not know it. Use it for most of my projects, research projects, personal tasks and automation of monotonous work.
Did and still prefer doing all my DSA in C, Even though Python is much more convinient, love the control and freedom C provides.
My Go-To for most of the statistical analysis, data visualization, and specific research projects requiring advanced statistical modeling. Learnt mainly through couple of Statistics courses but YouTube zindabaad!
Tools & Libraries
Hobbies & Interests
If you love running till you can't think anymore, let's run together!
I can't seem to put a label on the genre but it's broadly Philosophy, Psychology, self-help, Auto-Biographies and some niche topics. Reader? I'd love to know what you read. Thank you for this habit, Ma (Reema Khanna)
I love doing Analysis, Measure theory and Linear/Matrix Algebra. Currently exploring Differential Geometry and Riemannian geometry.
Learning from the Best
I mean, what is the point of life if you can't even listen to the most beautiful band in 'My Universe' under a 'sky full of stars' with 'Charlie Brown'. Trust me, It's 'Paradise', because it 'feels like I am falling in love'. Thank you for introducing it to me pa (Janesh Khanna) and Suhaan Khanna