- With more than 8 years of professional experience I have expertise in deploying AI Models for Large Scale VMS Products providing real-time analytics and deployed over Server, Cloud and Edge
- Thanks to my previous working on SQL Optimzer with proprietary contribution in Tree Parsing, Backpropagation and Tree Optimization Algorithm I precieve Neural Network functioning as different Tree Algorithms
- 🤠 I had lot of fun and learning while working intensively on Neural Network model parsing, converting to various frameworks, creating optimized files, implementing optimized network definition
- 💦 Last few years gave me an opportunity to contribute in every stage of AI-pipeline from data-collection to AI model deployment and creating products and libraries around AI models
- 🧠 Few hrs of week I spend in experimenting with new NN Architecture, playing with Optimiser Algorithm, trying different training strategies.
- 🧩 Currently I am actively working on extracting unique features from SSD without having additonal training
- 👷 I am open to contributing in AI it may be designing neural networks, formulating training strategies or writing new products for AI or with AI
- 📫 You can reach me on: ckhire91@gmail.com
| devchait.github.io |
Expertise: • Effective Problem Solving • Leading developers to extract there best • Architecting the VMS Product suite with real-time Analytics • Researching and selecting efficient tools for creating faster cost-effective AI/ML solutions • Designing Frameworks from scratch to leading them to full-fledge working suite • Selecting and formulating easy to contribute framework giving rise to automatic creation of new Products and automating existing workflows without explicit requirement for automation • Graph parsing algorithm, Tree Optimization Algorithm, Backtracking learning Algorithm • Working with PyTorch deep learning framework for design and implementation of CNN's for image recognition, classification, retrieval problems • Deploying NN models to highly optimized IE's including NVIDIA TensorRT, Intel Openvino, TFlite and Android NDK's • Optimizing end-to-end inference pipeline, Desigining new inferencing strategies for having larger throughput |
✔️🖋️ Simple CI/CD Setup with Gitea
✔️🖋️ Streamlined On-Premise Code Collaboration: Introducing Gitea - Your Effortless Open-Source GitHub Alternative
☄️🥇 AI developement evolution & AI/NN End-to-end Pipeline.
☄️🥇 Resposnible for design & implementation of ML workflow pipeline using DVC+Git.
☄️🥇 Multi-Faceted Inference Benchmark Framework
☄️🥇 Channel Improvement the impact of cuda base pre-processing
☄️🥇 Issue with Kalman and Deep Sort Tracking
☄️🥇 Simple Memory Efficient Adaptive Frame rate Tracker
☄️🥇 Custom Yolo Kernel for Faster Object Tracking
☄️🥇 Illusion of same operations on different neural network framework causing chaos
☄️🥇 Good Design Patterns and coding Practices leading to automatic formation of large AI ecosystem suite
☄️🥇 Simple Yolov5 Pre-processing CPU vs GPU
