Mandip Goswami

Principal Scientist — Acoustics, Audio AI & NVH @ Amazon FT & Robotics
mandipgoswami25@gmail.com  ·  Hugging Face  ·  LinkedIn  ·  GitHub

Summary

I am an acoustics scientist and machine-learning researcher focused on room acoustics, room impulse responses (RIRs), reverberant speech, and robust automatic speech recognition. My work centers on creating open, large-scale, well-documented datasets with reproducible evaluation harnesses — spanning simulated impulse responses, paired clean/reverberant speech corpora, industrial machine-sound anomaly detection, and interactive tools for acoustic analysis. Everything I release ships with machine-friendly metadata schemas, validation scripts, checksums, and reference baselines so that other researchers can reproduce and build on it.

Experience

Principal Scientist, Acoustics & Applied AI
Amazon — Fulfillment Technologies & Robotics
Apr 2025 – Present · Bellevue, WA
Senior Scientist, Acoustics & NVH
Amazon
Aug 2022 – Apr 2025 · Seattle, WA
Senior Acoustics NVH Engineer
Amazon — Worldwide Design & Engineering
Oct 2021 – Aug 2022 · Seattle, WA
Acoustics NVH Engineer
Amazon — Worldwide Design & Engineering
Nov 2019 – Oct 2021 · Greater Seattle Area
NVH Test Engineer — Vehicle Sciences
FCA Fiat Chrysler Automobiles
Jun 2016 – Nov 2019 · Auburn Hills, MI
Design & Development Engineer — Vehicle Interior R&D
Maruti Suzuki India Limited
Aug 2012 – Jul 2015 · Gurgaon, India

Selected Publications & Datasets

RIR-Mega: A Large-Scale Simulated Room Impulse Response Dataset for Machine Learning and Room Acoustics Modeling — arXiv (eess.AS, eess.SP), Oct 2025. arXiv:2510.18917
RIR-Mega-Speech: A Reverberant Speech Corpus with Comprehensive Acoustic Metadata and Reproducible Evaluation — arXiv, Jan 2026. arXiv:2601.19949
Whisper-RIR-Mega: A Paired Clean-Reverberant Speech Benchmark for ASR Robustness to Room Acoustics — arXiv, Feb 2026. arXiv:2603.02252
Acoustivision Pro: An Open-Source Interactive Platform for Room Impulse Response Analysis and Acoustic Characterization — arXiv, Feb 2026. arXiv:2602.12299
BeepBank-500: A Compact Synthetic Earcon / Alert Mini-Dataset for UI Sound Research — arXiv, Sep 2025. arXiv:2509.17277

Education

University of Cincinnati
M.S., Mechanical Engineering — Structural Dynamics & Vibrations
Maulana Azad National Institute of Technology (MANIT)
B.Tech., Mechanical Engineering

Honors & Awards

Certifications

Technical Skills

Psychoacoustics · Room Acoustics (FEM / BEM / Ray Tracing) · NVH Simulation · Audio Deep Learning (CNNs, Transformers) · Whisper Fine-Tuning · MFCC & Spectral Features · Sound Event Detection · Speaker Detection · Signal Processing · Python · Next.js · Hugging Face Ecosystem · Deep Neural Networks · Algorithms

Languages

English (Full Professional) · Hindi (Professional Working) · Japanese (Limited Working) · Assamese (Native / Bilingual)