open to research collaborations

Mihir Khatri

ai research

Working underneath the model

I'm a Computer Science student at Sardar Patel Institute of Technology, Mumbai, working on LLM inference efficiency, computer vision, and agentic systems. Most of what interests me lives one layer below the API — sparse sampling strategies, attention geometry, agent orchestration, and the exact place a system quietly breaks under load.

Published independent researcher (Ro-SVD, Zenodo 2026) and former AI/ML intern at Daten & Wissen. Currently building Zenso, a video engagement analytics platform, and Sentinel, a multi-agent harness for red-teaming LLMs across text, image, and video delivery.

9.35 / 10GPA, B.Tech Computer Science
35.1%KV-cache memory reduction, Ro-SVD
73.91%ROUGE-L win rate vs. uncompressed baseline

Current threads

Four lines of work, roughly in the order they started.

01 · published

Ro-SVD — Position-Aware KV Cache Compression (Zenodo, May 2026)

Position-aware truncated SVD on pre-RoPE key tensors for Grouped-Query Attention. Evaluated on 1,602 LongBench samples with Qwen2.5-7B-Instruct: 73.91% ROUGE-L win rate vs. uncompressed baseline, 35.1% KV-cache memory reduction. Custom CUDA de-rotation kernel with cuBLAS batched low-rank projection.

CUDAcuBLASLong Context
02 · inference efficiency

LLM Inference Optimisation — KV Cache via Low-Rank Approximation

End-to-end pipeline for inverse RoPE de-rotation, truncated SVD, and progressive online re-compression. 7-configuration sweep over 1,602 LongBench samples tracking ROUGE-L, perplexity, latency, and GPU memory. Explored Triton-based batched randomised SVD to cut the 29.5% Python-loop overhead.

PyTorchTritonCUDA
03 · in progress

Zenso — Neural Engagement Analytics

E(t) engagement scoring with hemodynamic lag correction. Six-module multi-agent architecture on LangGraph. In active development on private infrastructure (Kaggle/local), pending IP protection.

LangGraphFastAPIComputer Vision
04 · in progress

Sentinel — Cross-Modal LLM Red-Teaming

Multi-agent harness (LangGraph + OpenRouter/NVIDIA NIM) probing LLMs across five attack categories, extended to measure how jailbreak success shifts across text, image, and video delivery — the "modality gap." In active development on private infrastructure, pending IP protection.

LangGraphAI SafetyNVIDIA NIM

Selected work

SkillFlow AI ↗

Agentic peer skill-bartering platform built for the JPMC Agastya Foundation. Autonomous matching between learners and mentors, no manual moderation.

FastAPIAgentic
deployed — code ↗

Zenso

Neural engagement analytics platform built on sparse-frame video understanding and a six-module LangGraph agent architecture.

PythonLangGraph
active — private, pending IP protection

Sentinel

Multi-agent red-teaming harness for LLM safety, with cross-modal jailbreak measurement across five attack categories.

LangGraphOpenRouter
active — private, pending IP protection

CivicFlow

Automated form discovery and filling agent using Playwright and Claude Vision, with an RL feedback loop for adaptive form handling.

PlaywrightClaude Vision
complete

Automated Form-Filling Agent ↗

LLM agent that parses a target form's structure and generates Selenium automation to autofill it from a stored profile. Built for FAR AWAY 2026.

LangGraphSelenium
complete

Mineguard ↗

YOLO-based rockfall detection for safety-critical monitoring on limited annotated datasets.

YOLOComputer Vision
complete

Willovate Talent Hunt ↗

Production-grade ML service shipped under hackathon time constraints.

FastAPIDocker
complete

Transformer From Scratch

Multi-head self-attention, causal masking, positional encoding, and autoregressive decoding implemented from first principles.

PyTorch
complete

Track record

AI/ML Intern — Daten & Wissen Pvt. Ltd. 2025

Fine-tuned PaddleOCR models for license plate detection and recognition (~88% end-to-end accuracy). Evaluated U-Net/SAM/ResNet segmentation backbones for production latency–accuracy trade-offs. Ran Optuna hyperparameter sweeps.

Certifications

  • Hugging Face — Agents Course, Unit 1 Fundamentals Certificate
  • Hugging Face — NLP Course Certificate
  • Google Cloud Skills Boost — Agent Fundamentals
  • Google Cloud Skills Boost — Use Machine Learning APIs on Google Cloud (skill badge)
  • NVIDIA Deep Learning Institute — Agentic AI Explained
  • Docker Essentials: Container Networking

Get in touch