I'm a Data Scientist and ML researcher with 15+ years of experience (10+ in machine learning), currently at Ericsson Global, Bangalore. My work spans across spiking neural networks, large language models, AI applications in various domains such as web browser and telecom.
I am building practical ML systems while staying close to neuroscience-inspired computing, brain computer interfacing and more recently LLM applications. This github has some toy projects, some code for a few research papers, and some introductory lectures I gave.
- Spiking Neural Networks & Neuromorphic AI
- ML and LLM based Models for low-resource NLP (Bangla, Indic)
- AI + Mental Health: Mindful bots and guided tools
- Efficient ML: Tiny models, LoRA, and edge AI
- Telecom applications of AI (network optimization, ticket resolution)
- Pendulum Spiking Model: A biologically inspired spiking neuron model with Hebbian and STDP learning.
- BanglaScriptBot: A Bengali LLM-powered educational and writing tool.
- BiasNavigation: A tool to determine the amount of bias in news articles.
- MeditationRecommenderApp: A web-based app to recommend a type of meditation and give instructions based on your feelings.
- WebPageExtraction: A tool to extract relevant text (non-boilerplate, such as ads) from a web page. Used as input to a service such as Reader Mode in web browsers.
- A Hybrid Framework for Real-Time Data Drift and Anomaly Identification Using Hierarchical Temporal Memory and Statistical Tests. (2025)
- Identification of Inefficient Radios for Efficient Energy Consumption in a Mobile Network (IEEE 2024)
- PhD Thesis: Engineering a sequence machine through spiking neurons employing rank-order codes
- 3 patents filed in USPTO/WIPO related to AI in telecom: WO/2022/079724 FIRST NODE, AND METHOD PERFORMED THEREBY, FOR HANDLING ONE OR MORE CATEGORIES OF DATA, 20240196392 REDUCING INTERFERENCE IN A COMMUNICATIONS NETWORK and one unpublished (total 20 patents published, 6 granted in USPTO and WIPO)
- How can AI reduce time spent on categorizing CSPs' financial data
- Medium: Problems with the Generative AI Stack
If you're into spiking neurons, mindful tech, meditation gadgets, or interesting applications of LLMs, let's collaborate!