Resume

Via Aerea • Curriculum Vitae
Via Aerea • Par Avion 10c

Experience

Singularity Forge — Co-Founder & ML Engineer
Apr 2025 — Present
  • Leading SaaS development: RevQA (AI sales coaching), Logo Generation App, Voice-to-Text Notetaking App
  • Designing algorithms across multiple subnets in the Bittensor ecosystem for mining operations
  • Leading R&D in Satellite and Drone Technology for environmental mapping — DINO v3, sensor fusion for landmine detection and 3D modeling
Hover to flip
Singularity Forge
Co-Founder & ML Engineer
Postcard
Omega Labs — AI Engineer
Aug 2024 — Apr 2025
  • Any-to-Any Multimodal Models: Led development of seamless modality transformations, integrating image into speech-to-speech models
  • Diarization Pipeline: Built YouTube audio scraping/transcription system — 60,000+ row dataset on Hugging Face
  • Speech-to-Speech: Fine-tuned Mini-Omni, built multi-modality evaluation frameworks
Hover to flip
Omega Labs
AI Engineer
Postcard
Samsung R&D — Senior AI Engineer, Visual Intelligence
Dec 2021 — Dec 2024
  • Portrait Relighting: LoRA for Stable Diffusion, shipping on Samsung S25
  • OnlyMe Video Call: 60 FPS video instance segmentation for Galaxy S24
  • Selfie Camera Bokeh: Instance segmentation for S23, Flip4, Fold4 — 97.2 IoU, 60ms inference
Hover to flip
Samsung R&D
Senior AI Engineer, Visual Intelligence
Postcard
KAUST — Research Assistant — Vision CAIR Lab
Sep 2021 — Mar 2022
  • Developed 3D COMPaT dataset with Prof. Mohamed H. Elhoseiny — published at ECCV 2022 Oral Track
  • Part-material segmentation in 3D objects using branched DETR
Hover to flip
KAUST
Research Assistant — Vision CAIR Lab
Postcard
UC San Diego — Research Assistant — Embodied AI Lab
Jul 2021 — Sep 2021
  • Single-image to 3D shape reconstruction with Prof. Hao Su
  • Compositional learning framework using localized modules and substructure factorization
Hover to flip
UC San Diego
Research Assistant — Embodied AI Lab
Postcard
IISc Bengaluru — Research Assistant — Machine Learning Lab
Jul 2020 — Jul 2021
  • Unsupervised LiDAR reconstruction using GANs and Auto-encoders
  • Training techniques for Variational Autoencoders with global/local feature modeling using DCT
Hover to flip
IISc Bengaluru
Research Assistant — Machine Learning Lab
Postcard
Par Avion 20c

Education

ZHCET, AMU — B.Tech Computer Engineering
2016 — 2020

CGPA: 9.017/10.0 — Thesis: Analogy Generation

Servicio Postal 15c

Skills

Generative AI
Diffusion NetworksGANsAutoencoders3D Representation Learning
Languages
PythonC++SQL
ML Tools
PyTorch LightningPyTorchTensorFlowHugging Face
Vision
OpenCVPILAlbumentations
Web3
BittensorTAO MiningDecentralized Agents
Correo Aereo $1

Publications

Ujjwal Upadhyay, Yuchen Li, Habib Slim, Tezuesh Varshney, Ahmed Abdelreheem, Arpit Prajapati, Suhail Pothigara, Peter Wonka, and Mohamed Elhoseiny. "3DCoMPaT: Composition of materials on parts of 3D things."
ECCV 2022 (Oral)
Mansi Agarwal and Tezuesh Varshney. "Survey on Zero-Shot Learning and its application."
Via Aerea $5

Projects

3D Point Cloud Modeling

PyTorch implementation for 3D Point Clouds. Chamfer Distance and Earth Mover Distance from scratch. Generative Modeling on ShapeNet with Autoencoders and GANs.

Apery

Artistic image generation with heuristic algorithms. Transfer learning via VGG-19 for style transfer. Hill Climbing for image reconstruction using shape primitives.

Correo Certificado N.P.

Awards

Samsung Certificate of Appreciation S24, S23, Flip4, Fold4
HackData 2019 24-hr hackathon winner
Sir Syed Global Scholar Award Top 25 university-wide
NIOT SAVe 2019 4th of 128 teams, IIT-Madras
ACM-ICPC 2017-18 Best team from AMU
AMU-OSS Founder 150+ members
Google CodeJam 2018 Qualified
SnackDown 2018 Qualified