Arjun Balaji

MPA student with concentration in Data Science for Policy

Relevant coursework in Machine Learning, Statistics, Deep Learning, Signal Processing

Developed Unsupervised Models for Brain MRI Surface Registration in collaboration with Harvard Medical School at the Xu Lab.

Completing Bachelor Thesis on 3D Cardiac Segmentation at the MVAIT Lab.

Developed graph sampling techniques and anomaly detection algorithms for detecting money laundering in collaboration with NPCI at the VISTA Lab.

Researched memory-efficient techniques for Vision Transformers, achieving 8x memory compression with 96% accuracy.

Developed an LLM-driven, agent-based financial-wellness engine: an Insight agent derives persona + emotion cues, a planner–retriever loop (vector + rule search) selects interventions, and a shared memory layer preserves evolving user profiles for continuously adaptive guidance.

Open-sourced geospatial AI suite that fuses a Google Maps API tool for evaluating site proximity to essential services, a multimodal LLM pipeline that automates green-building certification tasks, and a Stable Diffusion + ControlNet workflow that synthesizes site-map datasets from edge and layout cues for model training.

Leveraged Power BI, Excel, SQL, and Python to automate data pipelines and build interactive dashboards that translate complex business datasets into clear, actionable insights, streamlining workflows and elevating data-driven decision-making.