Avinash Prabhu

Hello! I am Avinash Prabhu. I’m an ML Engineer at KLA, where I build vision-language models, CNNs, and tabular models for semiconductor defect inspection — from synthetic data generation to fine-tuning VLMs with SFT and GRPO to deploying them efficiently at scale with vLLM.

Previously, I did my MS in Computer Science at Georgia Tech, specializing in Machine Learning, and was a TA for Perception & Robotics and Deep Learning. I also interned with Apple’s  Computer Vision team, where I built a CNN-LSTM model for HomePod LED animation detection.

Before that, I completed my BTech (Honors) in ECE at IIIT-Hyderabad, where I was a research assistant at the Robotics Research Center, advised by Dr. Madhava Krishna, working on monocular 3D scene reconstruction. I also spent time at the Precog Lab, advised by Dr. Ponnurangam Kumaraguru, studying user behavior and platform moderation on Twitter using explainable ML.

Outside of work, I swim and play tennis. I’m always happy to chat over email or LinkedIn — especially if you’re working on fine-tuning VLMs for a specific use case, or squeezing more out of LLM/VLM inference.

News

Jun 2, 2024	Joined KLA as a Machine Learning Engineer.
Aug 20, 2023	Joined KLA as a Machine Learning Engineering intern.
May 21, 2023	Joined Apple  as a Computer Vision intern.
Aug 20, 2022	Started my MSCS at Georgia Tech, specializing in Machine Learning.
Jul 9, 2022	Graduated from IIIT-Hyderabad. Cannot be more grateful for the experiences and memories.
Feb 16, 2022	Our COVID-19 vaccination drive analysis is on Arxiv. Blog.
Dec 15, 2021	Our work on novel features for Twitter users published and presented at WI-IAT’21. Arxiv.
Dec 15, 2021	Our work on analysing restored Twitter users published and presented at WI-IAT’21. Arxiv.
Dec 14, 2021	Our work on monocular 3D reconstruction published and presented at ICVGIP’21. Arxiv.
May 3, 2021	Started my Computer Vision Internship at Robert Bosch.
Jan 19, 2021	Our Capitol Riots analysis on Twitter and Parler is on Arxiv.
Jan 1, 2021	Started my research at Precog.
May 1, 2020	Started my research at the Robotics Research Center.

Selected publications

Monocular Multi-Layer Layout Estimation for Warehouse Racks

Nigam, Meher Shashwat, Prabhu, Avinash, Sahu, Anurag and 5 more authors

In Proceedings of the Twelfth Indian Conference on Computer Vision, Graphics and Image Processing 2021

Abs

Given a monocular color image of a warehouse rack, we aim to predict the bird’s-eye view layout for each shelf in the rack, which we term as ’multi-layer’ layout prediction. To this end, we present RackLay, a deep neural network for real-time shelf layout estimation from a single image. Unlike previous layout estimation methods which provide a single layout for the dominant ground plane alone, RackLay estimates the top-view and front-view layout for each shelf in the considered rack populated with objects. RackLay’s architecture and its variants are versatile and estimate accurate layouts for diverse scenes characterized by varying number of visible shelves in an image, large range in shelf occupancy factor and varied background clutter. Given the extreme paucity of datasets in this space and the difficulty involved in acquiring real data from warehouses, we additionally release a flexible synthetic dataset generation pipeline WareSynth which allows users to control the generation process and tailor the dataset according to the contingent application. The ablations across architectural variants and comparison with strong prior baselines vindicate the efficacy of RackLay as an apt architecture for the novel problem of multi-layered layout estimation. We also show that fusing the top-view and front-view enables 3D reasoning applications such as metric free space estimation for the considered rack.
“I’ll Be Back”: Examining Restored Accounts On Twitter

Kapoor, Arnav, Jain, Rishi Raj, Prabhu, Avinash and 2 more authors

In IEEE/WIC/ACM International Conference on Web Intelligence and Intelligent Agent Technology 2021

Abs

Online social networks like Twitter actively monitor their platform to identify accounts that go against their rules. Twitter enforces account level moderation, i.e. suspension of a Twitter account in severe cases of platform abuse. A point of note is that these suspensions are sometimes temporary and even incorrect. Twitter provides a redressal mechanism to ‘restore’ suspended accounts. We refer to all suspended accounts who later have their suspension reversed as ‘restored accounts’. In this paper, we release the first-ever dataset and methodology 1 to identify restored accounts. We inspect account properties and tweets of these restored accounts to get key insights into the effects of suspension. We build a prediction model to classify an account into normal, suspended or restored. We use SHAP values to interpret this model and identify important features. SHAP (SHapley Additive exPlanations) is a method to explain individual predictions. We show that profile features like date of account creation and the ratio of retweets to total tweets are more important than content-based features like sentiment scores and Ekman emotion scores when it comes to classification of an account as normal, suspended or restored. We investigate restored accounts further in the pre-suspension and post-restoration phases. We see that the number of tweets per account drop by 53.95% in the post-restoration phase, signifying less ‘spammy’ behaviour after reversal of suspension. However, there was no substantial difference in the content of the tweets posted in the pre-suspension and post-restoration phases.
Efficient Representation of Interaction Patterns with Hyperbolic Hierarchical Clustering for Classification of Users on Twitter

Karandikar, Tanvi, Prabhu, Avinash, Tulasi, Avinash and 2 more authors

In IEEE/WIC/ACM International Conference on Web Intelligence and Intelligent Agent Technology 2021

Abs

Social media platforms play an important role in democratic processes. During the 2019 General Elections of India, political parties and politicians widely used Twitter to share their ideals, advocate their agenda and gain popularity. Twitter served as a ground for journalists, politicians and voters to interact. The organic nature of these interactions can be upended by malicious accounts on Twitter, which end up being suspended or deleted from the platform. Such accounts aim to modify the reach of content by inorganically interacting with particular handles. These interactions are a threat to the integrity of the platform, as such activity has the potential to affect entire results of democratic processes. In this work, we design a feature extraction framework which compactly captures potentially insidious interaction patterns. Our proposed features are designed to bring out communities amongst the users that work to boost the content of particular accounts. We use Hyperbolic Hierarchical Clustering (HypHC) which represents the features in the hyperbolic manifold to further separate such communities. HypHC gives the added benefit of representing these features in a lower dimensional space – thus serving as a dimensionality reduction technique. We use these features to distinguish between different classes of users that emerged in the aftermath of the 2019 General Elections of India. Amongst the users active on Twitter during the elections, 2.8% of the users participating were suspended and 1% of the users were deleted from the platform. We demonstrate the effectiveness of our proposed features in differentiating between regular users (users who were neither suspended nor deleted), suspended users and deleted users. By leveraging HypHC in our pipeline, we obtain F1 scores of upto 93%.
Capitol (Pat)riots: A comparative study of Twitter and Parler

Hitkul, , Prabhu, Avinash, Guhathakurta, Dipanwita and 9 more authors

2021
Co-WIN: Really Winning? Analysing Inequity in India’s Vaccination Response

Karandikar, Tanvi, Prabhu, Avinash, Mathur, Mehul and 3 more authors

2022