Iram Kamdar

About Iram Kamdar

Hi! I’m Iram, an MS Data Science student at Columbia University specializing in applied machine learning, LLM engineering, and scalable retrieval systems. I’ve worked across Dolby Labs, WebAI, and Columbia’s Computational Imaging group on projects spanning audio fingerprinting, vector DB optimization, LoRA fine-tuning, and clinical NLP.I love turning ML ideas into end-to-end reliable systems, from data pipelines and model evaluation to inference optimization and deployment. Recently, my focus has been on agentic AI workflows, vector search architectures, and making open-weight models work in low-latency environments.Let’s connect!

Experience

1. Data Scientist

   Dolby Laboratories

   Sep 2025 — Present · NY, US

   Built a contrastive-learning audio fingerprinting model with CNN encoders + FAISS retrieval to identify distorted audio segments across 100K+ samples; improved sequence matching and temporal modeling under real-world distortions.• Redesigned evaluation logic for variable segment durations (1–10 s) to boost retrieval robustness. Automated hit-rate testing, improving top-1 accuracy under noise.

Education

• SRM IST Chennai

  Bachelor of Technology - BTech, Computer science with specialisation in artificial intelligence and machine learning

• Columbia University

  Master of Science - MS, Data science

• SRM IST Chennai

  Bachelor of Technology - BTech, Computer Science