Neham Jain

Research Scientist · Meshy AI

Neham Jain

I work on generative models and 3D.

Researcher · meshy.ai

Say hello CV Scholar GitHub LinkedIn

Photo of Neham Jain

About

Before Meshy, I built human motion-capture pipelines at Meta Reality Labs and worked on neural rendering at Adobe Research. I earned my MS in Robotics at Carnegie Mellon, where my thesis explored reconstructing scenes through smoke. These days most of my time goes into scaling 3D models and data and, unglamorously, looking at the data. The single best debugging tool I know is scrolling through a thousand random training samples.

Off the clock I'm cooking something slightly too ambitious (results vary), watching football (the round-ball kind), staying up way too late for F1 qualifying, or planning the next hike. Lowkey always down to debate any of the above.

Selected work

  1. 2026

    Meshy-7

    I helped build Meshy-7, Meshy's image-to-3D foundation model. Go try it out!

  2. 2025

    Embody3D

    Meta's large-scale multimodal dataset of human motion. I built the multiview capture pipelines, pose and hand-tracking models, and QA filters behind the release. Paper

  3. 2025

    SmokeSeer 3DV 2026

    Seeing through smoke: fuses RGB and thermal video with 3D Gaussian splatting to remove smoke and reconstruct the scene behind it. My MS thesis at CMU. Videos · Thesis

  4. 2024

    Relightable splats for 3D foundation models

    At Adobe Research I wrote CUDA kernels that replace spherical harmonics with neural features, and scaled training to 32 nodes. Per-asset time went from 6 minutes to 30 seconds.

  5. 2024

    Counterfactual Explanations of Visual Recommender Systems WWW 2024

    Explaining visual recommendations through small changes to image embeddings, with language explanations grounded using CLIP. Work from my research internship at Adobe.

  6. 2022

    Towards Realistic Underwater Dataset Generation and Color Restoration ICVGIP 2022

    Bridging synthetic and real underwater imagery with image-to-image translation, creating a more realistic training dataset for underwater color restoration. Research at IIT Madras.

Experience

  • Meshy AI Research Scientist2026 –
  • Meta Reality Labs Research Engineer2025 – 26
  • Adobe Research Research Engineer Intern2024
  • Carnegie Mellon MS Robotics · GPA 4.112025
  • IIT Madras BTech Electrical Engineering2023

Full experience →

Notes

  • May 2026Talk at Voxel51's Best of 3DV
  • Mar 2026Joined Meshy to work on 3D foundation models
  • Winter 2025SmokeSeer accepted at 3DV 2026
  • Fall 2025Embody3D released by Meta Reality Labs

Contact

Always happy to talk shop. The quickest way to reach me is nehamjain2002@gmail.com.