Aahan Rembersu

Building AI agents and LLM orchestration pipelines as part of a cross-functional squad at Apexon, working alongside engineering and domain interns under the mentorship of a senior architect. • Developing LangChain-based generative AI pipelines for real-world enterprise applications, contributing to a reusable internal Apexon asset. • Performing data preparation, exploratory data analysis, and statistical modeling to support predictive and causal inference solutions. • Building UI prototypes using Streamlit to demonstrate AI application capabilities and gather iterative user feedback. • Collaborating with data scientists and engineering partners to design, build, and deploy data science solutions, communicating results clearly to non-technical stakeholders.

Developed and experimented with deep learning models using Python, Jupyter, and PyTorch within agile, governance-aligned workflows. Applied linear regression and CNNs on 50K+ sample real-world datasets in R, improving accuracy by up to 12% via preprocessing, hyperparameter tuning, and robust evaluation. Optimized pipelines with batched loaders, mixed precision, and early stopping, cutting end-to-end experiment time by 30–40% and supporting scalable ML solutions.

Bachelor of Engineering in Computer Engineering at PICT with GPA 8.65/10. Led as Sports Representative on the Student Council, increasing sports participation ~20% and coordinating intercollegiate events. State-level tennis athlete with National Gold & Silver; team titles at MIT-WPU SUMMIT ’23, MIT-ADT VISHWANATH 2024, and FLAME Kurukshetra 2024. Built Marathi Grammatical Error Correction system using IndicBART/Varta-T5/Marathi T5 on a 100K+ pair dataset; evaluated with BLEU