We at Paperplane have onboarded 3,000+ doctors who have completed 120k+ appointments for 600k+ patients.- Built and deployed a conversational agent over a database of marketing content and results of a diabetes instrument company to judge and plan their broadcasts to their customers.Challenges solved: Chat history on limited contextwindow of LLM, Output parsing for graphs, .json and .csv, Embeddings and Vector Database for data index, Function calling.- Developed from scratch a Graphical Recommendation Engine which has assisted doctors to write 5,000 Digital Prescriptions under 15 seconds via AI nudges without much clinical data to start with.- Deployed Transformer Network with Attention-blocks to detect intent behind the patient’s query in natural language over WhatsApp and respond with the result in real time (under 100ms).- Trained and deployed Sequence-to-SequenceTransformer (T5) to generate promotional marketingcontent from keywords for doctors.I trained this model on synthetic data generated by a bigger model and I quantized weights for inference under 5 seconds.- Built Data Pipelines and Embeddings for ContinualLearning of the models.- I co-led the Design team for the first iteration of theproduct, working on the UX.- Transitioned into the product role where I was responsible for the B2C side of business and was the touchpoint for customers and tuning the product alongside my team via user feedback. Improving onboarding speed and hence MAU by 3x.

Devised mathematical formalism for Cognitive models for Collective intelligence, metaphors and Decentralised Science.

Heading the Special Interest Group curated for individuals with a keen interest in Artificial Intelligence and Cognition.

I created a project for Network Traffic Analysis through different protocols like TCP, ARP, UDP, etc.It was done for further analysis over the Institute's network.