Shubham Vashishth | Data Scientist

Hi, I'm Shubham Vashishth.

A

Self-driven, quick starter and a passionate AI/ML enthusiast with a strong background in developing innovative solutions for diverse industries.

About

Shubham Vashishth is a seasoned machine learning engineer with over 1.5+ years of experience in the IT industry, specializing in developing AI-driven solutions across multiple industries. A graduate in Computer Science Engineering from Guru Gobind Singh Indraprastha University and currently pursuing his Master’s in Computer Science from the prestigious University of Stuttgart. This blend of academic rigor and professional experience empowers Shubham to solve complex, real-world problems with cutting-edge technology.

His notable work includes spearheading the development of an AI-based transportation solution for real-time vehicle analysis, using optical flow technology to detect wrong-way vehicles and optimizing traffic flow through adaptive signal control, resulting in a 15% reduction in traffic jam time over the rush hours.

Alongside his work in AI-based transportation solution, Shubham contributed to the development of Optimus, a production scheduler tailored for the pharmaceutical industry. Optimus optimizes production processes, driving efficiency and productivity across multiple pharmaceutical plants.

Driven by a passion for AI and problem-solving, Shubham continuously seeks out new challenges to expand his expertise. His diverse experience across domains, coupled with his dedication to innovation, makes him a valuable contributor in the fields of machine learning and artificial intelligence.

If you’re interested in collaborating on an exciting project, feel free to reach out!

Experience

Ripik AI

Machine Learning Engineer

Improved the design and development of Optimus, a sophisticated production planner tool tailored for the pharmaceutical industry, leveraging linear optimization algorithms for optimal scheduling.
Developed and implemented algorithms, specifically designed to accommodate the constraints and requirements of plant production teams.
Implemented robust optimization techniques that resulted in a 10% increase in production efficiency across six different Pharma plants. Utilized LLMs to curate comprehensive monthly reports of the entire schedule, presenting data in a user-friendly format with visualization graphs and analytics.
Engineered a sophisticated compliance tracking system that meticulously compares actual plant performance against the Optimus-generated schedule over a monthly production cycle. This feature enabled in-depth performance analytics, identification of deviations, and facilitated continuous process optimization.
Provided training and support to plants end-user, facilitating smooth adoption and maximizing the tool's impact on production workflows.
Tools: Python, Statistical Data Analysis, Data Engineering, Knowledge Engineering, Data Visualization, Data Structures

Apr 2024 - Aug 2024 | Noida, India

Constems AI

AI Software Intern -> AI Engineer

Contributed in a project to develop an AI-based model for real-time vehicle data analysis on roads, deployed successfully on Indian roads in collaboration with NHAI.
Individually led the conceptualization, design, and execution of a comprehensive pipeline, specializing in the detection of vehicles moving in the wrong direction on roadways. Innovatively harnessed optical flow technology to establish a sophisticated solution, showcasing technical prowess in creating an end-to-end pipeline for precise identification and reporting of misaligned traffic movements.
Implemented machine learning models for predicting traffic bottlenecks and adaptive traffic signal control algorithms based on reinforcement learning techniques, resulting in a 15% reduction in overall travel time.
Developed a robust product recognition system for unstructured store layouts, improving accuracy in identifying and pricing products by 25% and reducing instances of mislabeling.
Engineered a state-of-the-art product identification model that demonstrated a 97% accuracy in discerning diverse brands and their respective variants, showcasing a keen eye for detail.
Served as a mentor to junior interns, offering comprehensive guidance on standard operating procedures to ensure a thorough understanding and seamless integration into the workflow.
Tools: Python, Computer Vision, Image Processing, OpenCV, Keras, Tensorflow, PyTorch, YOLO.

Apr 2023 - Apr 2024 | Remote, India

Cron Labs

Data Scientist Intern

Developed an automated trading system using Python and AlphaTrade API for data analysis and execution of trades, implemented machine learning algorithms for identifying profitable trades and optimized trade execution which was faster by 47% than previous versions.
Build an Object Detection model using YOLOV5 to detect various features of local kirana/medical shops like assets, products, machinery etc.
Built and maintained data pipelines using AWS Data pipeline and managed data lake architecture on S3 for data storage and retrieval.
Tools: Python, Computer Vision, Data Engineering, Machine Learning, AWS, SQL

Dec 2022 - Mar 2023 | Bengaluru, India

Add Innovations Pvt Ltd.

Machine Learning Intern

Expertise in debugging and troubleshooting current code.
Developed algorithms for surface inspection and detection, including image enhancement, defect detection and pattern recognition.
Wrote code to preprocess, annotate, and augment image datasets for training computer vision models.
Implemented deep learning models for image classification and segmentation using Tensorflow frameworks.
Tools: Python, Image Processing, OpenCV, Keras, Tensorflow, PyTorch, YOLO.

Sep 2022 - Dec 2022 | Greater Noida, India

Google Cloud

Cloud Trainee(Mentorship program)

Created, managed and deployed cloud resources to kubernetes on cloud platform.
Performed foundational Data AI/ML tasks, hands on qwiklabs and weekly assignments.
Tools: Google cloud platform(GCP)

Aug 2021 - October 2021 | Remote, India

Projects

                
Dash Cam Plate And Facial Recognition
                  Dash cameras with AI-driven license plate extraction and facial recognition for improved traffic surveillance and security.
                
AccomplishmentsCustom deep neural networks with transfer learning, optimized using TensorFlow and GPU acceleration.
Deployed CNNs for real-time object detection using PyTorch, R-CNNs, and non-maximum suppression.
Engineered OCR with RNNs, bidirectional LSTMs, and fine-tuned pre-trained models for license plate recognition.
Implemented Siamese networks with contrastive and triplet loss functions for facial recognition, fine-tuning for accuracy.

Blur Detection
                  Engineered an image quality assessment system that detects and quantifies blurriness in an image.
                
AccomplishmentsCreated a CNN model for detecting image blur with 95% accuracy, using diverse datasets and Laplacian variance.
Serialized data and model for reproducibility and seamless integration.
Applied normalization for better training convergence and interpretation.

Stock Price Analysis and Prediction Using LSTM and RNNs 
                  Built an LSTM-based model to forecast stock prices, utilizing historical data from yfinance, optimizing with scaling and windowing, and presenting trends with Matplotlib for strategic insights.
                
AccomplishmentsBuilt an LSTM neural network to predict Google stock prices.
Used yfinance to download and analyze historical stock data.
Applied data scaling and windowing to optimize inputs for the LSTM model.
Created charts with Matplotlib to provide insights into historical and predicted prices.

Gym Bro 
                  Engineered a squat counter leveraging MediaPipe and OpenCV for advanced pose detection and precise squat counting.
                
AccomplishmentsBuilt a squat counter app using MediaPipe and OpenCV for real-time pose detection and squat counting.
Developed pose estimation with MediaPipe and image processing with OpenCV to track and count squats.