AI/ML Engineer specializing in Computer Vision & Generative AI

Solving complex engineering problems through ML, computer vision, and generative AI — NYU & IIT Madras alum | Intel award-winning intern | Research Assistant at NYU

View my projects View my experience Download resume
A portrait of Manoj Srinivasan
Google Scholar

About Me

👋 Hello! I'm Manoj, an AI/ML engineer who thrives on solving complex challenges at the intersection of computer vision, generative AI, and 3D rendering. My passion isn't just research — it's turning cutting-edge ideas into practical, impactful solutions.

I’m a recent Master’s in Computer Engineering graduate from New York University and an alumnus of IIT Madras. Currently, I’m a research assistant at the NYU Video Lab (advised by Prof. Yao Wang), developing diffusion-based solutions for 3D scene refinement and reconstruction.

I love bringing applied AI projects to life. During my internship at Intel, I received an Excellency Award for developing novel vision-based automation tools for integrated chip design validation. I’ve also contributed to fast-growing startups like Preimage and GalaxEye Space, where I worked on 3D reconstruction pipelines and satellite image super-resolution, respectively.

On the academic side, I worked on "U2NeRF: Unsupervised Underwater Image Restoration and Neural Radiance Fields" as part of my master’s thesis, focusing on 3D-consistent unsupervised image restoration for underwater scenes. This work was later published at the ICLR ’24 Tiny Papers conference, held in Vienna.

🚀 I’m always excited to connect with teams pushing the boundaries of AI for engineering, robotics, and visual intelligence. If that sounds like your world, I’d love to connect and explore opportunities to build something impactful together.

Work Experience

A timeline of my professional roles and contributions.

Graduate Research Assistant, NYU Video Lab

Aug 2025 – Present (New York City, NY)

NYU Logo
  • Implemented a pipeline integrating 2D video diffusion priors with 3D Gaussian Splatting (3DGS), achieving a 19% LPIPS improvement in novel-view synthesis under sparse input conditions.
  • Enhanced 3D view consistency by formulating view generation as a temporal continuity task, integrating camera-pose embeddings with diffusion-guided latent features across viewpoints.
  • Exploring mesh registration for human poses using learned skinning methods such as SMPL, to improve 3D consistency.
Python PyTorch Computer Vision 3D Gaussian Splatting Diffusion Models HPC Slurm Meshlab

Graduate Engineering Intern, Intel Corporation

Jun 2024 – Aug 2024 (Santa Clara, CA)

Intel Logo
  • Developed a computer vision framework for automated detection of IC package design violations, emulating manually-performed inspection heuristics.
  • Optimized computational efficiency using segmentation models and OpenCV's morphological algorithms, reducing detection pipeline runtime by 85% (from >4 hours to <30 minutes).
  • Honored with Intel’s Impact Award for strong productivity and delivering high-quality solutions in a short timeframe.
Python Computer Vision Image Processing HuggingFace OpenCV Xpedition VBScript

Graduate Research Student, Indian Institute of Technology, Madras

Aug 2022 – May 2023 (Chennai, India)

IIT Madras Logo
  • Proposed U2NeRF, a fully self-supervised transformer-based framework for joint underwater image restoration and neural 3D reconstruction, embedding physics-informed light modeling into the NeRF pipeline.
  • Leveraged a disentangled representation of underwater degradations, which includes scene radiance, global illumination, and scattering maps, to enable accurate color and structure recovery in the absence of ground-truth supervision.
  • Introduced patch-level rendering to address limitations of pixel-wise NeRF, enabling improved local spatial context modeling for underwater image restoration.
  • Achieved state-of-the-art results on the newly curated Underwater View Synthesis (UVS) benchmark across 12 calibrated scenes, with +11% perceptual similarity and +4% restoration quality over prior methods.
  • Developed as part of Master’s thesis at IIT Madras and later published at ICLR 2024 (Tiny Papers Track).
Python Transformers Neural Radiance Fields (NeRF) 3D Reconstruction PyTorch Distributed Data Parallel (DDP)

Machine Learning Intern, Preimage

Sep 2022 – Dec 2022 (Bangalore, India)

Preimage Logo
  • Adapted a transformer-based multi-view stereo (MVS) pipeline for dense 3D geometry reconstruction from aerial (drone) imagery, optimizing performance for sparse-view and large-scale outdoor scenes.
  • Improved feature representation using an adaptive feature pyramid network (FPN) that leverages sinusoidal embeddings conditioned on scene-specific depth bounds, increasing reconstruction accuracy by ~10% in challenging outdoor scenarios.
  • Deployed large-scale reconstruction experiments on Azure VMs, leveraging AWS S3 for dataset management and PyTorch Lightning with CUDA for efficient distributed training.
Python 3D Reconstruction Transformers PyTorch Lightning PyTorch Azure AWS S3

Machine Learning Intern, Asilla Japan

May 2022 – Jul 2022 (Remote)

Asilla Japan Logo
  • Enhanced an abnormal activity detection pipeline for surveillance systems by fine-tuning the quantization mechanism used to learn a latent “normal action” dictionary and flag deviations as anomalous behavior.
  • Reduced runtime latency by ~15%, enabling deployment on 20+ real-time CCTV feeds at Hankyu Nishinomiya Gardens Mall, Japan.
Python Computer Vision Machine Learning Bash SSH Git/Github scikit-learn OpenCV

Image Processing Intern, GalaxEye Space

Dec 2021 – Jan 2022 (Chennai, India)

GalaxEye Space Logo
  • Built a super-resolution neural network to upsample low-quality remote-sensing data in the form of SAR images, along with a generative model to predict RGB optical images from the super-resolved SAR output.
  • Conducted experiments on cross-public datasets, leading to increased super-resolution quality even at scales up to 16x.
Python Generative Adversarial Networks (GANs) Computer Vision PyTorch Git/Github Docker Weights & Biases

Education

New York University (NYU), Tandon School of Engineering

Sep 2023 – May 2025 (New York, NY)

NYU Logo
  • M.S. in Computer Engineering
  • Cumulative GPA: 3.93/4.0

Indian Institute of Technology Madras (IITM)

Aug 2018 – Jul 2023 (Chennai, India)

IIT Madras Logo
  • B.Tech in Mechanical Engineering & M.Tech in Robotics (Dual Degree)
  • Cumulative GPA: 3.51/4.0
  • Minors in Computing, Artificial Intelligence & Machine Learning

Featured Projects

Here are some of the projects I'm proud of. Feel free to check them out.

Project U2NeRF Project U2NeRF Preview

U2NeRF (ICLR '24)

Self-supervised transformer-based framework achieving joint underwater image restoration and 3D reconstruction, embedding physics-informed light modeling for realistic view synthesis.

3D Reconstruction PyTorch Neural Radiance Fields
Project Website →
3D Gaussian Splatting Enhancer

3DGS Enhancer for Sparse Views

Diffusion-guided framework integrating 2D video priors with 3D Gaussian Splatting to improve sparse-view reconstruction, enhancing 3D view consistency through pose-conditioned feature alignment.

3D Computer Vision Video Diffusion PyTorch
Project Report →
Counterfactual Image Generation Using Text Guidance Project

Counterfactual Image Generation Using Text Guidance

Text-guided counterfactual generation framework combining CLIP and Stable Diffusion to inpaint high-confidence regions, achieving fine-grained attribute manipulation while preserving over 90% of original content.

Computer Vision Stable Diffusion CLIP
Project Report →
Satellite Image Super-Resolution

Satellite Image Super-Resolution

GAN-based system that fuses SAR and optical images to turn low-resolution satellite data into sharper, more detailed outputs using cross-domain feature alignment.

Computer Vision PyTorch Generative Adversarial Networks
Github Repository →
Implicit 3D Surface Mesh Reconstruction

Implicit 3D Surface Mesh Reconstruction

Implicit ML-framework that reconstructs 3D geometry from point clouds, learning a continuous function (occupancy or SDF) over space.

3D Reconstruction PyTorch Signed Distance Function
Github Repository →
Project Steel Defect Project Steel Defect Preview

Steel Surface Defect Detection

Deep-learning model that detects and segments defects on steel surfaces from images, generating pixel-wise masks for precise localization and handling various defect types.

Semantic Segmentation PyTorch U-Net
Github Repository →

Honors & Recognition

Key accomplishments and recognitions received throughout my career.

Intel Impact Award

Intel Corporation

August 2024

Awarded the "Impact Award" during my Graduate Engineering Internship for delivering high-quality work ahead of schedule and demonstrating exceptional productivity.

Published Author - ICLR '24

ICLR Conference, Vienna

March 2024

Research paper titled "U2NeRF: Unsupervised Underwater Image Restoration and Neural Radiance Fields" accepted and published at the prestigious International Conference on Learning Representations (ICLR), Tiny Papers track.

Bronze Medal – Inter-IIT Tech Meet 2023

IIT Kanpur, India

February 2023

Awarded Bronze (3rd place) in the Chandrayaan Moon Mapping Challenge organized by ISRO, representing the IIT Madras contingent at the 2023 Inter-IIT Tech Meet.

All India Rank 1467 - JEE Advanced

IIT Kanpur, India

June 2018

Secured an All India Rank of 1467 in the highly competitive JEE Advanced, placing among the top 0.1% engineering aspirants in India.

National Talent Search Exam (NTSE) Scholarship

NCERT, India

June 2016

Awarded the prestigious NTSE Scholarship, granted to Grade 10 students ranking in the top 0.4% nationwide.

Get In Touch

I'm currently open to new opportunities and collaborations.. If you have a project in mind or just want to say hi, feel free to reach out!

Say Hello