Available for Research & AI Collaborations

Applied Scientist | 1x Entrepreneur | Spatial AI (3D & 2D) | Medical AI & Neuroimaging | 7+ Years Experience
Publications5 Papers
Experience7+ YearsAI & Startups
Core Focus3D Spatial AINeuro & Medical AI
Research & Industry AI Showcase
6 Core Domains
3D Brain Tractography & Point Clouds
IIT Mandi
3D Spatial AI

3D Brain Tractography & Point Clouds

HVAC Vector Path & Geometry Estimation
Attentive.AI
Vector AI

HVAC Vector Path & Geometry Estimation

CADtnet Ultrasound Breast Cancer Triage
xtLytics LLC
Medical AI

CADtnet Ultrasound Breast Cancer Triage

ArchGlyphOCR Blueprint Plan Parsing
Attentive.AI
Document OCR

ArchGlyphOCR Blueprint Plan Parsing

Invoice OCR Table Contouring
xtLytics LLC
Document Layout

Invoice OCR Table Contouring

PointGPT Streamline Foundation Models
IIT Mandi
3D Spatial AI

PointGPT Streamline Foundation Models

About & Vision

Applied Scientist, Researcher, and Entrepreneur with 7+ years of experience across 3D Spatial AI, Medical Imaging, Document Layout OCR, and Reinforcement Learning. Founded Tensr.AI (Incubated at SIIC IIT Kanpur, HTIC IIT Madras, IIT Mandi Catalyst). Currently pursuing M.Tech (by Research) at IIT Mandi under Dr. Arnav Bhavsar and Dr. Aditya Nigam, developing SOTA 3D point cloud transformer models for white matter tractography and neuro-radiology extensions.

Industrial & Entrepreneurial Experience

Attentive.AI

Attentive.AI

Current Role
Gurgaon, IN
Jun 2025Present
Research Engineer 2
  • Architected a vector path tokenization framework (PDF → SVG) as a pretraining objective for zero-shot symbol detection and vector path annotation, increasing SOTA macro-F1 by 7%.
  • Trained and shipped DEIMv2 (DINOv3 backbone) for multi-class vector path segmentation and geometry dimension estimation, reducing total annotation time (TAT) for HVAC to 35% of original duration.
  • Designed, built, and deployed ArchGlyphOCR: a dense plan text-region proposal + prediction pipeline outperforming Google DocAI & Gemini 3.1 Pro on callout extraction while eliminating API costs.
IIT Mandi

IIT Mandi

Mandi, IN
Feb 2023Jun 2025
Researcher
  • Designed foundational models for 3D streamlines using PointGPT style architectures with novel tokenization, dual-masking pretraining, and multi-embedding representation learning for WM tract segmentation.
  • Built and deployed SlicerTracto, an open-source 3D Slicer extension for brain MRI tractography (reconstruction, tracking, segmentation, visualization); deployed at PGI Chandigarh Hospital.
  • Published 5 papers at ICPR'24 (TractoEmbed, TractRL-former), ISBI'25 (TractoGPT), and ISBI'26 (TrackletGPT, TractRLFusion), outperforming Harvard Medical School (TractCloud) and SCIL Canada (FIESTA).
StatusNeo

StatusNeo

Gurgaon, IN
Aug 2022Mar 2023
Lead Data Scientist
  • Led the Cortex AI team under Dr. Shailesh Kumar for Reliance Jio's TelOS MLOps ecosystem across multiple verticals.
  • Built a programming language & framework-agnostic ML library via Protobuf schemas and ONNX serialization.
  • Mapped comprehensive model ontologies, aligning 6 Scikit-Learn models and PyTorch libraries into a unified cross-framework schema (sklearn ↔ PyTorch ↔ TensorFlow).
Gamezop

Gamezop

Gurgaon, IN
Jun 2021Jun 2022
Senior Data Scientist
  • Created ClickVision: a Markov Decision Process (MDP) deposit prediction model over large-scale clickstream trajectories using RL transition matrices.
  • Reduced wasted offers and user churn by 62%, significantly boosting deposit conversion rates.
  • Shipped downstream UI and product changes based on model trajectory predictions.
Tensr.AI

Tensr.AI

Noida, IN
Feb 2020Jun 2021
Founder & CEO
  • Built pyCardiograms (LSTM-based SCG waveform prediction, wave annotation across ECG/PPG/SCG).
  • Scaled team to 15 members; DST & DPIIT Government certified startup.
  • Secured incubation at SIIC IIT Kanpur, HTIC IIT Madras, and IIT Mandi Catalyst; ranked #3 at Himalayan Startup Trek 2019.
  • Designed and prototyped Teresa Band, a proprietary health wearable capturing mechanical heart vibrations.
xtLytics LLC

xtLytics LLC

Noida, IN
Jan 2019Jan 2020
Data Scientist
  • Invoice OCR: Built GCP Vision + Tesseract + Faster-RCNN pipeline to extract and structure tabular content into CSV/XLSX via geometric bounding box heuristics.
  • CADtnet: Developed edge TFLite model (VGG-19, MobileNet, ResNet) for mobile breast cancer triage on ultrasound images, deployed in LMICs.

Skills & Technology Stack

3D Spatial AI (Point Clouds)
PyTorch & PyTorch3D
Medical AI & Neuroimaging
DEIMv2 & DINOv3
3D Slicer Extension (SlicerTracto)
Reinforcement Learning (Gym)
Document OCR (ArchGlyphOCR)
Transformers (GPT / BERT)
Open3D & Fury.gl
ONNX & Protobuf Schemas
Python
C++
Scikit-Learn
TFLite Edge Deployment
Research Milestones

Peer-Reviewed Publications

Acceptances and key contributions across 3D Spatial AI, Brain MRI Tractography, and Reinforcement Learning.

Peer-Reviewed Papers:

"My research focuses on 3D spatial AI, point cloud sequence modeling, and non-invasive neuroimaging (diffusion MRI white matter tractography), outperforming established benchmarks from Harvard Medical School (TractCloud) and SCIL Canada (FIESTA)."

Accepted at IEEE ISBI 2026 (London, UK): TrackletGPT: A GPT architecture for White Matter Segmentation (Rank A). B-Spline sub-segment tokenization for 3D point clouds.

3D Spatial AI
B-Spline Tokenization
GPT Transformers
PyTorch3D

Accepted at IEEE ISBI 2026 (London, UK): TractRLFusion. Reinforcement Learning policy fusion networks for Diffusion MRI tractography.

Reinforcement Learning
Fusion Networks
Diffusion MRI
PyTorch

Accepted at ISBI 2025 (Houston, TX): TractoGPT. Dual-masked point cloud GPT model. Outperformed SCIL Canada's FIESTA benchmark.

3D Spatial AI
Dual-Masked Pretraining
GPT Transformers
SCIL Benchmark

Accepted at 27th ICPR 2024 (Kolkata): TractoEmbed. Modular multi-embedding CNN/dVAE/PointNet framework. Outperformed Harvard Medical School's TractCloud.

Representation Learning
PointNet / dVAE
Harvard Benchmark
Neuroimaging

Accepted at 27th ICPR 2024: Tract-RLformer. Supervised + RL 2-stage policy refinement. Outperformed SCIL Canada's Track2Learn, DeepTract, and PFT.

Reinforcement Learning
Decoder Transformers
SCIL Benchmark
PyTorch
Open Source Software & Tools

Open Source Software & AI Tools

Standalone extensions, 3D Slicer tools, and multimodal spatial editing frameworks.

Open-source 3D Slicer desktop extension enabling Neuro-Radiologists to visualize, track, and segment white matter tracts from Diffusion MRI scans in a standalone GUI app with optional on-prem/cloud GPU compute offloading. Deployed at PGI Chandigarh.

3D Slicer
Medical AI
Dipy / Scilpy
PyTorch
Python

Multimodal Text & Sketch-to-3D point cloud generation and interactive editing framework using SDFusion + LLaMA/GPT continuous feedback loops for granular region-specific 3D spatial manipulation.

Multimodal 3D AI
SDFusion
LLaMA / GPT
Open3D
PyTorch3D
Honors & Recognition

Awards & Incubation Grants

National incubation grants, fellowships, and startup recognitions.

DST & SIIC IIT Kanpur

NIDHI-EIR Fellowship Grant

2021 - 2022
DST & SIIC IIT Kanpur

Awarded National NIDHI Entrepreneur-in-Residence fellowship grant by Department of Science & Technology (DST) government of India, incubated at SIIC IIT Kanpur for Teresa Band biomedical wearable development.

HTIC IIT Madras & GDC Foundation

HTIC IIT Madras Incubation & GDC Fellowship

2020 - 2021
HTIC IIT Madras & GDC Foundation

Pre-incubated at Healthcare Technology Innovation Centre (HTIC) IIT Madras; completed Gopalakrishnan Deshpande Foundation customer discovery program with 85+ clinical interviews.

IIT Mandi Catalyst

IIT Mandi Catalyst Incubation Grant

2020 - 2021
IIT Mandi Catalyst

Selected for startup incubation and prototype grant support at IIT Mandi Catalyst following the Himalayan Startup Trek.

IIT Mandi

Rank #3 — Himalayan Startup Trek 2019

Sep 2019
IIT Mandi

Ranked 3rd out of 279 competing startups at Himalayan Startup Trek 2019 for Tensr.AI cardiovascular signal assistant.

Specialized Credentials

Professional Certifications

Verified specialization and coursework credentials in deep learning, reinforcement learning, and PyTorch.

Research & News

Recent Updates

Ongoing milestones across research, startup, and industry journey.

TrackletGPT & TractRLFusion Acceptances

London, UK

Two papers accepted at ISBI 2026: TrackletGPT (Rank A) and TractRLFusion, pushing white matter tract segmentation into new reinforcement and transformer paradigms.

TractoGPT Accepted at ISBI 2025

Houston, Texas

Paper accepted at 23rd ISBI 2025: TractoGPT for non-invasive neuroimaging analysis.

TractoEmbed & Tract-RLformer Accepted at ICPR 2024

Kolkata, India

Papers accepted at the 27th ICPR: TractoEmbed and Tract-RLformer.

GitHub
LinkedIn
X