Kanishk Varshney

AI Researcher

Bengaluru, Karnataka, India9 yrs 2 mos experience

Highly StableAI ML Practitioner

Key Highlights

Innovated scalable AI systems for real-world impact.
Reduced manual annotation costs by 40% in AI projects.
Achieved industry benchmarks in satellite imagery processing.

Stackforce AI infers this person is a Geospatial AI and Machine Learning expert with a focus on scalable systems.

Contact

Skills

Core Skills

Mlops & Scalable Ai SystemsGenerative Ai & Deep LearningOptimizing PerformanceComputer Vision

Other Skills

Cloud AI & Distributed ComputingCross-functional Team LeadershipEnglishProject ManagementPublic SpeakingPyTorchPythonResearchTeamwork

About

🚀AI Researcher | ML Innovator | Scaling AI for Real-World Impact I specialize in Generative AI, Deep Learning, and Geospatial Intelligence, building scalable AI systems that drive real-world impact. 🔹 Key Innovations: - Geospatial AI & Super-Resolution – Optimized large-scale satellite imagery processing with Transformer-based models. - AI for Disaster Management – Accelerated post-disaster assessments by 30%, enhancing real-time response. - AI for Driver Safety (Netradyne) – Designed Drowsiness Detection & Face Recognition AI, securing 10,000+ drivers. - MLOps & Automation – Led ATLAS and VYOM, reduced manual annotation costs by 40%, and built scalable ML pipelines. 🔹 Patents & Research 📌 Digital Image Registration (2024) – AI-powered precision for geospatial data 📌 AI for Road Infrastructure (2022) – Advancing intelligent mapping systems Passionate about advancing AI research, optimizing ML systems, and bridging deep learning with classical AI. Let’s connect and innovate!

Experience

Aidash

3 roles

Data Science Manager

Apr 2024 – Dec 2024 · 8 mos

Innovated Annotation Systems: Designed and implemented ATLAS, a scalable annotation platform that reduced manual labeling efforts by 40%.
Developed MLOps Platform: Created Vyom, a Kubeflow-based ML platform improving AWS GPU training efficiency by 40%.

MLOps & Scalable AI SystemsPyTorch

Staff Data Scientist

Apr 2023 – Apr 2024 · 1 yr

Led AI models analyzing 360-degree panoramic imagery for U.S. DoTs and government projects.
Built ODIN: End-to-End ML Deployment System, reducing model rollout time by 40%.
Streamlined real-time inference workflows for AiDASH, integrating AWS, GitHub, and Jenkins CI/CD.

Generative AI & Deep LearningMLOps & Scalable AI Systems

Senior Data Scientist

Feb 2021 – Apr 2023 · 2 yrs 2 mos

Developed AI-driven disaster response systems, reducing damage assessment time by 30% post-Hurricane Ida.
Achieved sub-1.5m satellite image alignment using GAN-based co-registration algorithms, improving geospatial data precision.
Pioneered 4x super-resolution for satellite imagery, setting a new industry benchmark with SwinIR, Transformers & Stable Diffusion.

Generative AI & Deep LearningMLOps & Scalable AI Systems

O9 solutions, inc.

Senior Software Engineer

May 2020 – Feb 2021 · 9 mos · Bangalore Urban, Karnataka, India

Optimized Python-based forecasting models, increasing execution speed by 4x for more accurate demand predictions.
Enhanced plugin execution pipelines, boosting data load/unload speed by 10x for enterprise applications.

PythonCloud AI & Distributed ComputingOptimizing Performance

Netradyne

2 roles

Senior Research Engineer

Promoted

Jul 2018 – Apr 2020 · 1 yr 9 mos · Bengaluru, Karnataka, India

Developed AI-powered drowsiness and yawn detection systems to enhance driver safety, reducing fatigue-related incidents by 15%.
Engineered a real-time facial recognition system for improved security, resulting in a 0.5% reduction in False Acceptance Rate (FAR).

PythonCloud AI & Distributed ComputingComputer Vision

Research Engineer

Nov 2016 – Jun 2018 · 1 yr 7 mos · Bengaluru, Karnataka, India

Developed AI-Powered Seat Belt Detection system with 98%+ accuracy using HOG, Adaptive Thresholding & Gamma Correction.
Created Image & Video Annotation Tool based on VATIC for scalable dataset generation.
Implemented real-time eye closure tracking, blink rate monitoring, and pupil size evaluation for improved drowsiness detection accuracy.

PythonCloud AI & Distributed ComputingComputer Vision