Amit Singh Rajawat

I am Amit Singh Rajawat, a Software Engineer at Kloudspot Inc. , passionate about Machine Learning, Deep Learning, and Computer Vision.

I hold a B.Tech in Information Technology (AI & Robotics) from MITS, Gwalior. I have gained skills in DBMS, ML, OS, Probability, CN, Robotics, CV, and NLP. I was awarded the IIT Madras Pravartak Undergraduate Fellowship 2023-24, a prestigious award given to only 72 students across India.

In my current role, I have designed algorithms for face recognition technology evaluation and built models on various datasets using ResNet-200 architecture. My contributions have led to significant improvements in face recognition tasks, achieving notable ranks in global evaluations. As a former intern at Kloudspot Inc., I created a high-performance computer vision inferencing framework and implemented gRPC for effective client-server communication.

I have also gained valuable experience during my project internship at the Indian Space Research Organisation (ISRO), where I worked on developing a vision-based satellite pose estimation system using the HRNet algorithm.

My research contributions include publications on machine learning strategies, programmed face monitoring systems, and recognition of Parkinson’s ailment using machine learning procedures.

More information can be found in my CV. Feel free to email me at imamitsingh958@gmail.com for research discussions or collaborations.

Experience

Software Engineer
Kloudspot Inc., Bengaluru, India | June 2024 - Present

Designed a Face Recognition Algorithm in C++ for 1:1 Verification and 1:N Identification, achieving a global rank of 277 in 1:1 Verification and a global rank of 59 in 1:N Identification in the NIST-Face Recognition Vendor Test (FRVT).
Trained models using ResNet-200 with CBAM, incorporating knowledge distillation and reinforcement learning, improving accuracy by 13%.
Developed an FRS Analysis Tool using clustering techniques and a FRS Registration Tool in Python—implementing detection, landmark processing, and recognition for multi-pose registration and real-time tracking with visualizations to monitor passenger movements.
Developed and maintained scalable data pipelines for real-time facial recognition data and integrated CI/CD practices for automated deployment and continuous improvement of deep learning models in production environments.

Software Engineering Intern
Kloudspot Inc., Bengaluru, India | Feb 2024 - May 2024

Created a high-performance computer vision inferencing framework in C++ leveraging GStreamer, OpenVINO, and DLStreamer, capable of managing multiple models efficiently, 9% improvement in processing speed compared to the Python-based implementation.
Implemented gRPC for effective communication between client and server, along with GStreamer RTSP Server for seamless real-time video broadcasting which supported high-quality video transmission with minimal packet loss.
Innovated pre-processing and post-processing workflows that refine the flow of video data and improve the accuracy of model predictions.

Machine Learning Intern at Indian Space Research Organisation - ISRO (LOR)
Indian Space Research Organisation - ISRO, Bengaluru, India | Dec 2023 - Jan 2024

Worked on developing a vision-based satellite pose estimation system using the HRNet algorithm, which enhanced landmark detection accuracy for ISRO’s upcoming missions. The system leveraged HRNet's multi-resolution feature, enabling precise detection of key landmarks on satellite images in challenging conditions, such as varying lighting, occlusions, and satellite orientations.
Engineered the integration of deep landmark regression and nonlinear pose refinement techniques to accurately determine satellite orientations, training the landmark regression model on extensive satellite image datasets to capture spatial relationships.

Amazon ML Summer School 2023
Amazon | Sep 2023 - Oct 2023 | Apprenticeship

Skills: Casual Interface · Reinforcement Learning · Deep Neural Networks (DNN) · Dimensional Reduction · Unsupervised Learning · Supervised Learning · Sequential Learning · Probabilistic Graphical Models.

ML Research Intern
Madhav Institute of Technology & Science, Gwalior | May 2023 - July 2023

Collaborated with the faculty, conducted experiments, analyzed data, and refined models to improve the accuracy on real-world projects. Wrote Research Papers on Iris flower species and Face monitoring system and publish them in an IEEE Conference.

Backend Developer
Punama Innovation | March 2023 - April 2023

Worked on live Media Network website.

Full Stack Developer
Madhav Institute of Technology & Science, Gwalior | Jan 2022 - Dec 2022

Created a dynamic website for registering students in NEC Courses, equipped with both an Admin panel and a Faculty panel. This website is designed to serve the college's registration needs efficiently. (Certificate Link)

Web Development Intern
Sparks Foundation Network (Link) | Sept 2021 - Oct 2021

Make a Donation website to Donate Money for Charity. Generates receipt and send a copy to donator and one copy at organization which conduct the donation camp.

Publications

Categorization of IRIS Flower using Different Machine Learning Strategies
A. Rajawat, A. Manjhi and A. Srivastava
2023 IEEE International Conference On Electrical, Electronics, Communication and Computers (ELEXCOM), Roorkee, India, 2023
[PDF]

Design and Analysis of Programmed Face Monitoring System
A. Manjhi, A. Rajawat and A. Srivastava
2023 IEEE International Conference On Electrical, Electronics, Communication and Computers (ELEXCOM), Roorkee, India, 2023
[PDF]

Recognition of Parkinson’s ailment by using various machine learning procedures
Rajawat, A.S., Srivastava, A.
Curr Psychol (2024)
[PDF]

On Heart Disease
(Communicated in Journal)

On Phishing Website
(Communicated in Journal)

Technical Skills

Programming Languages: C/C++ (Proficient), Python (Proficient), JavaScript, HTML/CSS, LaTeX, Shell, MySQL
Developer Tools and Frameworks: Git, GitHub, Linux, Docker, VS Code, ONNX, AWS, CI/CD, MLOps
Libraries: Pandas, NumPy, PyTorch, TensorFlow, Scikit-learn, Flask, OpenCV, Transformers, MXNet, Keras, LLMs
Coursework: DBMS, Machine Learning and Optimization, OS, DSA, CN, Robotics, Computer Vision, NLP, AI

My Projects

StoryCraftAI - Infinite Possibilities in Every Story | Python, TensorFlow, Keras, NLP, Pandas, Huggingface, CI/CD (Link)

Developed and optimized a suite of deep learning models for story generation using advanced NLP techniques. Improving model accuracies after training: GRU increased by 13.46%, LSTM by 33.53%, Bidirectional-LSTM by 25.32%, and Bidirectional-GRU by 9.91%.
Integrated an interactive Flask-based web interface for real-time story generation and deployed the solution via CI/CD on Hugging Face Spaces.

GestureSymphony - Control video playback using your hands | Python, OpenCV, MediaPipe, Pygame, Flask, NumPy, Bootstrap (Link)

Engineered a real-time hand gesture recognition system using MediaPipe Hands and OpenCV for detection and control commands, demonstrating strong computer vision and ML integration skills. Dynamic playlist management to control video playback (pause, play, next, previous).
Enhanced system efficiency by implementing lazy loading and containerizing with Docker; established a robust CI/CD pipeline for continuous integration and deployment on cloud platforms like Render, ensuring scalable and reliable real-time performance.

Snake Game Using Reinforcement Learning With Neural Network (Link)

Enhanced gameplay using Q-learning, improving the agent's performance through trial and error with state-space representation and boolean features.
Visualized agent performance and evolving strategies with animated plots for a deeper understanding of decision-making.

Text Classification With TensorFlow (Link)

Developed a high-performance text classification model for sentiment analysis, and document categorization, leveraging advanced techniques such as NLP and deep learning algorithms. Optimized model performance through hyperparameter tuning, feature engineering, and cross-validation.
Deployed the optimized model using TensorFlow, demonstrating advanced machine learning skills in text-based applications.

Rainfall Prediction Using ML (Link)

Designed a neural network model to predict rainfall with 95% accuracy, improving predictions through feature engineering and model optimization.
Provided actionable insights to support agricultural planning, water resource management, and disaster preparedness.

Movie Recommendation System Using ML With GUI (Link)

Developed a personalized recommendation system using collaborative filtering and adaptive logic based on user preferences, trends, and collaborative behavior, with genre-based filtering and weighted ratings for more accurate and relevant recommendations tailored to each user's search.
Deployed on a Flask-based GUI for real-time movie search and personalized recommendations, enhancing user experience with different features.

Hand Gesture Recognition using OpenCV (Link)

Trained a Temporal Convolutional Network (TCN) on 12,000 RGB videos to learn efficient temporal modeling in image sequences for recognizing hand gestures, enabling downstream music control tasks such as pause, play, and next. Integration into a real-time application pipeline.
Achieved 92% accuracy by pre-processing images with background removal techniques, using color-based segmentation and SIFT for feature extraction.

Novel Engaging Course Project - Live Project (Link)

Developed and deployed a novel and engaging live project as part of the curriculum, aligning with the objectives of the National Education Policy (NEP) 2020, bringing a first-of-its-kind initiative to Madhya Pradesh.
Successfully deployed the project, which has been running for 1.5 years, attracting thousands of registrations from learners across various backgrounds, fostering a dynamic and interactive learning environment.

Multiplayer Pong Game (Link)

Created an engaging multiplayer Pong game using socket.io, HTML, CSS, Node.js, and JavaScript, demonstrating expertise in web development, real-time communication, and interactive user interfaces.
Integrated APIs to enhance game functionality, enabling seamless data exchange and dynamic gameplay features, showcasing proficiency in integrating external services into the project.

Omicron Sentiment Analysis Using ML (Link)

Contributed to the development of a scalable and robust system, integrating data preprocessing, model training, and deployment, resulting in an efficient sentiment analysis tool to monitor and track public opinion surrounding the Omicron variant.
Executed a sentiment analysis project, leveraging machine learning algorithms to analyze social media data and classify sentiment associated with mentions of the Omicron variant, enabling rapid understanding of public sentiment during a global health crisis.

Phishing Website Analysis Using ML (Link)

Developed a machine learning-based project for phishing website analysis, employing advanced algorithms and feature extraction techniques to accurately detect and classify fraudulent websites, enhancing online security and user protection.
Implemented data preprocessing, feature engineering, and model training to build a robust classification system that effectively distinguishes between legitimate andalicious websites.

Payment Gateway Integration — Internship Project (Link)

Successfully completed an internship project focused on payment gateway integration, developing a robust solution that generated payment receipts and sent them via email to both the payer and recipient, ensuring a seamless and transparent payment experience.
Implemented payment gateway APIs, securely handled financial transactions, and leveraged email APIs to automate the generation and delivery of payment receipts.

Stock Price Prediction Using ML (Link)

Developed a machine learning model for stock price prediction using historical stock data and relevant features, contributing to improved investment strategies and financial decision-making.
Conducted extensive data preprocessing, feature engineering, and model evaluation to optimize the prediction system's performance, resulting in accurate and reliable stock price forecasts.