- Working as a consultant after having completed an internship to help improve the CAC (Customer Acquisition Cost) system and further automate the process of constructing an effective marketing strategy based on the past and current performance metrics of an online advertisement- Working to minimize the need for human intervention in the current AI systems

- Contributed significantly in developing the Smart Audience AI system to automatically analyse customers’ product marketing requirements and build specific audience cohorts to market the products to- Contributed new ideas and solutions to the existing Campaign AI system, a proprietary deep neural-network based AI model capable of re-distributing marketing budgets among advertisements for a product based on 100+ different metrics to minimise marketing costs by up to 30%- Developed an automated CAC (Customer Acquisition Cost) analysis system capable of extracting insights from pre-published product advertisements, identifying factors which do not work well from a marketing standpoint, and suggesting strategies to maximise product marketability and minimise CAC

- First-authored a research paper titled 'Cause-Augmented Shared Fusion-Based Multimodal Sarcasm Explanation Generation' in which we proposed a novel Generative AI model, CASH, that can analyse sarcastic multimodal social media posts and explain the sarcasm in them- Developed a new technique to combine multimodal semantic information and introduced improvements to the techniques used by the pre-existing state-of-the-art- Beat the performance of the previous SOTA by 3% on average and proved the efficacy of each component of the model in achieving this performance via thorough comparative analyses and ablation studies- Also beat a cutting-edge multimodal LLM (LLaVA-v1.6-Mistral-7b) by 24% across all metrics to demonstrate the superiority of CASH over existing LLM-based systems - Submitted the paper to EMNLP 2024 (2nd largest international NLP conference)