Shubham Pateria

 Leading multiple funded projects in complex long-horizon multi-agent Reinforcement Learning and privacy-preserving Federated Learning.
 Strong expertise in Deep Learning, Reinforcement Learning, Federated Learning, Python, and Pytorch.
 Developed and successfully licensed a deep multi-agent learning and self-organizing neural networks hybrid system achieving above 90% success rates in large-scale defense forces automation tasks. Technology transferred to DSO Laboratories.
 Built domain knowledge of Federated Learning from scratch without prior experience and launched multiple ongoing projects on federated human activity recognition, distributed data classification, distributed clustering, and time-series domain adaptation.

Introducing AI to creatives such as designers, writers, creative directors, artists, etc.

Digital mental health startup. Discontinued.

gomaargo.com (mobile).

https://dblp.org/pid/186/7303.html

Working on algorithms for autonomous decision-making via hierarchical and deep reinforcement learning. Robotic navigation. AI agents for games.

Reinforcement learning refers to the training of machine learning models to make a sequence of decisions that optimize an objective in the future.

• Developed a multi-agent hierarchical reinforcement learning approach for robot coordination for Search & Rescue (in simulation), as part of the STE-NTU Corporate Lab project ‘Joint Situation Awareness and Cooperative Reinforcement Learning’.

• Wrote the first-ever comprehensive survey covering the whole breadth of the Hierarchical Reinforcement Learning research up to the recent years.

• Developed novel hierarchical learning algorithms using topological/graphical memory and goal-decomposition for long-horizon decision-making, which can be applied in tasks such as robot navigation, manipulation, and AI-based game playing.

Served as a teaching assistant for the following courses:

1. CZ2001: Algorithms, Semester 2, 2018.
2. CZ3003: Software System Analysis and Design, Semester 1, 2019.
3. CZ2001: Algorithms, Semester 2, 2019.
4. CZ1015: Introduction to Data Science, Semester 1, 2020.
5. CZ3004: Multidisciplinary Design Project (MDP), Semester 2, 2020.

R&D Contribution: Novel methods for optimizing power consumption and visual quality of the Samsung smartphone display modules. The work led to two issued patents.

1. Worked with Display technology teams in SRI-B and Samsung HQ responsible for board bring-up and device driver upgrades critical for the successful commercial launch of Samsung smartphones in the worldwide market (Galaxy S4-variants, A7, Tab4, and other mid-tier smartphone variants).

2. R&D Contribution: Machine learning for power-efficient and fault-tolerant sensor management system for Smart Home IoT sensors with minimal error (average 0.24$^{\circ}$C) in temperature prediction. Published in IEEE IACC.

Created didactic content for virtual Chemistry lab as part of the Lab-on-a-Tab project.