Dr. Prathosh A.P.
Assistant Professor · Dept. of ECE · Indian Institute of Science, Bengaluru
Co-Founder & Head of AI · LatentForce.ai
कुर्वन्नेवेह कर्माणि जिजीविषेच्छतं समाः।
एवं त्वयि नान्यथेतोऽस्ति न कर्म लिप्यते नरे॥ Īśāvāsya Upaniṣad
Assistant Professor · Dept. of ECE · Indian Institute of Science, Bengaluru
Co-Founder & Head of AI · LatentForce.ai
कुर्वन्नेवेह कर्माणि जिजीविषेच्छतं समाः।
एवं त्वयि नान्यथेतोऽस्ति न कर्म लिप्यते नरे॥ Īśāvāsya Upaniṣad
I am an Assistant Professor in the Department of Electrical Communication Engineering at the Indian Institute of Science (IISc), Bengaluru, where I lead the Representation Learning Lab (RepLab). My research focuses on how machine learning systems can learn and be corrected reliably when data is scarce, noisy, or biased.
Before IISc, I was an Assistant Professor at IIT Delhi (2017–2021) and held research positions at Xerox Research Centre India and Philips Innovation Labs. I received my Ph.D. from IISc in 2015 and B.E. from SJCE Mysore in 2011. I also hold a concurrent appointment as Visiting Faculty at the Centre for Brain Research (CBR), IISc.
I co-founded LatentForce.ai (IISc-FSID incubated, USD 1.7M seed funding) and Cogniable, an autism intervention platform acquired by Frontera Inc., USA in 2024. I serve on the Committee for Responsible AI, Government of Karnataka.
IISc, September 2021 – present.
Learning reliably from noisy or scarce data using submodular selection, DPP-based attribution, and hypothesis testing. Applied to LLM fine-tuning, drug response prediction, and anomaly detection.
Concept removal with formal guarantees across GANs, diffusion models, and discriminative classifiers. Covers selective forgetting and black-box output filtering.
Recovering symmetry groups — discrete and continuous Lie groups — directly from data, with invariance built into the architecture.
Neural operators in the complex Fourier domain for improved OOD generalisation. Applied to PDE solving and real-time air quality forecasting (R² > 0.95).
Reducing annotation requirements in medical image analysis — from bounding-box supervision down to fully self-supervised diffusion-based methods.
₹25K/month top-up salary + ₹2.5 Lakhs/year research grant.
AI infrastructure for coding agents. The flagship product, LatentGraph, indexes entire codebases into queryable knowledge graphs for structurally grounded context retrieval. Raised USD 1.7M in seed funding (Ideaspring Capital, Yali Capital) in December 2025.
AI-driven platform for early autism intervention, deployed at AIIMS Delhi, MAMC, and NIMHANS. Served over 1,000 children across India using video analytics and behaviour assessment. Recipient of the UNICEF XTC EdTech Global Prize and the India Innovation Growth Programme award.