Jul 2021 – Sep 2024 · 3 yrs 2 mos

• LLM-based external knowledge chatbots: Worked on creating an e-commerce retail consumer-facing LLM-based chatbot, automating the complete product ordering journey.
• QnA RAG chatbots: Created a dashboard API where users can do QnA with their documents via LLMs and vector databases like ChromaDB and FAISS.
• TM Forum Jio Async API Hackathon Challenge: Was a part of Jio’s Hackathon team of 5 people, where we developed TMF AIVA, a multilingual LLM-based QA chatbot for Telecom use cases. Came in 2nd place in the same.
• LLM Dashboard: Developed a complete end-to-end dashboard for users to finetune their custom Q&A dataset on any LLM and interact with the finetuned LLM directly in our chatbot interface.
• IndicLens Dashboard: Developed a complete end-to-end dashboard for users to use our novel IndicLens pipeline and power their product catalogue with our novel multimodal and multilingual search capabilities.
• Multilingual Hate Speech Classification: Ran multiple experiments for generating rich embeddings for hate speech tweets data and optimal semantic understanding of the text. Work is currently submitted for a patent.
• Domain Prediction: Developed an ML-based end-to-end pipeline for classifying a query into multiple domains, using multiple data processing, cleaning and feature engineering techniques. Currently deployed in MyJio app and increased search CTR by 15.2%.
• Indic Transliteration and Translation: Developed an end-to-end transliteration and translation pipeline and REST APIs for major Indian regional languages for all domains of Jio’s e-commerce platforms, increasing media and e-commerce search CTR by 12.4%.
• Query Category Prediction: Developed an end-to-end pipeline and REST APIs for recognising categories to which the query belongs at every category level using NLP feature engineering and modelling techniques. Increased product search page CTR by 24%.

Jan 2021 – Apr 2021 · 3 mos

• Developed a PoC for NLP-based extraction of product attributes from product descriptions for fashion products and tested the same with different algorithmic methods.

Aug 2020 – Dec 2020 · 4 mos

• Worked on various modules towards the development of search engine pipeline for a fashion e-commerce platform along with other data science analysis, research and experimentation tasks related to the pipeline.

May 2020 – Aug 2020 · 3 mos

• Worked on search query optimization for AJIO Fashion and developed a knowledge graph API scalable to multiple e-commerce domains, titled J.A.N.K.I (Just a New Knowledge Graph).

Apr 2020 – Aug 2020 · 4 mos

• Worked under Prof. Amit Sheth on the topic of multimodal analysis for toxic tweet classification, using natural language processing, knowledge graphs and basic image processing techniques along with literature review and experimentation for the same.

Jul 2019 – Jul 2019 · 0 mo

• A work from home internship based on an introduction to Image Recognition models and traffic light detection system for the project: AyeAI Autonomous Ambulance.

May 2019 – Jul 2019 · 2 mos · Mumbai, Maharashtra, India

May 2018 – Jun 2018 · 1 mo · Mumbai, Maharashtra, India