My arXiv paper on generating spoken esports-style gameplay commentary with a vision-language model and text-to-speech — no game-specific code required.
Research
I am a researcher specializing in machine learning for gaming, on-device AI, and ML model optimization. Currently at Qualcomm working on ML on edge for gaming (2025–present), previously at Activision on ML-based anti-cheat systems (2022–2025). I hold an MS in Electrical Engineering from the University of Washington and an MS in Data Science from Plaksha University.
Research Interests
- Machine Learning for Gaming
- On-device AI / Edge ML
- ML Model Optimization
- Computer Vision
- Natural Language Processing
- IoT and Embedded Systems
Research Profiles
| Platform | ID / Link | Purpose |
|---|---|---|
| Google Scholar | PH5u7jIAAAAJ | Publications & citations |
| ORCID | 0009-0002-1103-3441 | Researcher identifier |
| OpenReview | ~Mathew_Varghese1 | ML/AI conference reviews & submissions |
| Microsoft CMT | mathewvarghesemanu@gmail.com | Conference management & paper submissions |
Publications
-
AI Level of Detail: Distance-Aware ML Model Precision Selection for Real-Time Human Motion Prediction in Games (2026)
Mathew Varghese · PRICE Workshop, SIGGRAPH 2026 Technical Workshops
ACM DL · arXiv · Slides -
Call of Duty: Warzone Caldera OpenUSD Data Set (2024)
Koen Vroeijenstijn, Mathew Varghese, Xav de Matos, Senwhaa Lim, Michael Vance, Natasha Tatarchuk
GitHub Repository -
IOT and PLC Based Home Automation System with PV Inverter (2019)
Ameer Fayiz, Athira Prasad, Mathew Varghese
Full publication list: Google Scholar profile.
Research Posts
AI Level of Detail: Distance-Aware ML Model Precision Selection for Real-Time Human Motion Prediction in Games
AI Level of Detail (AI LOD): using distance-aware neural-network quantization as a level-of-detail axis for real-time NPC motion prediction. FP32/FP16/INT8 ONNX tiers routed by camera distance — up to 9.79x faster inference with no perceptible quality loss.
SCI Unit Test – Carbon Footprint of Deep Learning Models
Data PII Tokenisation
In association with HRS
3d Printing Cost Estimation Using Machine Learning
In association with UC Berkeley
IoT and PLC Based Home Automation System with PV Inverter
A smart home automation system with the solar inverter presented as the final thesis at engineering college