Hi everyone!
I am Siddhi and my passion for building machine-learning solutions has made me pick machine-learning engineering as a career. I have been working with machine learning, deep learning, computer vision, and machine learning operations for the last 5 years. Currently, I am just a student with a lot of time to work on my repo. You can check out my projects here.
My first job was as a data scientist in one of the subsidiaries of a prestigious fintech company in Nepal called extensodata. At extensodata, I mostly tangled with huge structured fintech data ranging from banks to e-wallets. I began realizing that data science is a huge field and can get very vague.
I wanted to specialize in computer vision and joined Leapfrog, where I worked on some image segmentation and object tracking projects. After working for a year, I decided to continue my studies and joined the graduate program at the University of South Dakota.
- Languages and Scripts: Python, Bash, C/C++
- Frameworks and Libraries: Pandas, Scikit-learn, Numpy, Seaborn, Plotly, Scipy, Django Rest Framework, Keras(with Tensorflow), TF Lite, Tensorflow TensorRT, Pytorch, Darknet (For YOLO), OpenCV, Tesseract, Ultralytics.
- IDE: NVim, Jupyter Notebook, Pycharm, VSCode
- Database: MySql, PostgreSQL, MongoDB
- VCS: Git, Github, Bitbucket, Gitlab
- Cloud services: (AWS) S3, EC2, Lambda
- Containerization, and orchestration: Docker, Docker-Compose.
- Collaboration: JIRA, Trello, Slack
- Methodology: Scrum, Kanban
- ETL: Pentaho, Airflow, Dragster
- Messaging Broker: RabbitMQ
- Big Data Technology: Apache Spark, HDFS, PySpark (SQL, MLLib)
- Visualization: Microsoft PowerBi, Apache Superset, Python Libraries (Matplotlib, Seaborn, Plotly, Dash)
- ML-Ops: Weights and Biases, MLflow, Tensorflow-Serve, ClearML.
- Hardware: Jetson, Oak cameras (luxonis), Raspberry Pi.
- Operating System: Linux, Windows, MacOS.
i8 Labs Inc, Mountain View, California, USA
- Enhanced mAP score by 26% and improved training efficiency by establishing in-house annotation standards and leading computer vision model training and deployment using custom YOLO models with the Ultralytics framework for accurate people, bicycle, and car counting across diverse visual conditions.
- Implemented ClearML for MLOps and deployed IoT solutions using Balena, managing deployment to devices with OAK, Raspberry Pi, and ESP32 controller and enabling efficient tracking of datasets, models, and experiments to improve reproducibility and deployment automation.
- Conducted rigorous internal audits of deployed models to ensure accuracy, enhancing client satisfaction and supporting their goal of securing funding to maintain and expand trails.
MSc. Computer Science at University of South Dakota (Vermillion, USA)
- Took a break after a successful career in machine learning to pursue further academic studies.
- Published a paper at the IEEE AI Conference and co-authored a book on machine learning.
Leapfrog Technology, Kathmandu, Nepal
- Lead and mentored the AI/ML team for project delivery.
- Lead end-to-end client requirement elicitation process.
- Defined & developed standard ML practices.
- Used deep-learning frameworks such as darknet, OpenCV, and Tensorflow TRT to train & evaluate YOLO models for object detection.
- Build, Deploy and Maintain statistical, ML, and Deep learning using standard ML\MLOps frameworks such as MLFlow.
- Model tuning and optimization focused especially on deep learning models for embedded devices (NVIDIA Jetson Developer Toolkit).
- Worked on a multi-object tracking project using quantized YOLO tiny models for object detection, & deepsort for tracking.
- Worked on a prototype for a calorie estimator by segmenting the items on a plate using Masked RCNN.
- Experience working with human-computer interaction.
- Worked as team manager for the AI team.
Extensodata Pvt. Ltd, Kathmandu, Nepal
- Use and development of Data Architectures.
- Explanatory Data Analysis (EDA) in SQL as well as Jupyter notebooks.
- Using big data tools such as Hadoop, Spark, Hive, etc to manage huge volumes of data effectively.
- Data visualization using python libraries (seaborn, Matplotlib) and other third-party tools such as PowerBi & Apache Superset.
- Using various machine\deep learning models in spark (MLLib) as well as python (Sci-kit Learn, Keras).
- Using Pentaho and spark for extraction, transformation, and loading data from raw data (files, database, HDFS, hive) to required data architecture.
- Study feasibility, pros, and cons of machine learning and statistical models.
- Query optimization in a relational database (Mysql) for quicker data analysis.
- Writing automation scripts for various purposes (such as ETL, web scraping, etc) using python and Linux shell scripts.
- Studying the application of machine learning models in the banking domain.
- Generating and studying relevant using different feature engineering techniques (such as custom and quartile binnings, combining multiple features) in bank-specific data.
- Building prototype machine learning models on an ad-hoc basis as well as deployable backend data structures.
- Writing stored procedures and scripts to generate various reports from source data for UI consumption.
- Mentoring interns, trainees, and Junior members of the Team
Introduction to Machine Learning in Production
Coursera (July 2022)
https://coursera.org/verify/AQYAFFTRKJW9
Optimize TensorFlow Models For Deployment with TensorRT
Coursera (June 2022)
https://coursera.org/verify/K343P63ZCMNR
Speak Like a Pro: Public Speaking for Professionals
Udemy (June 2022)
UC-9c99a21f-818b-43fa-9f06-34f457356a6d
Deep Learning Computer Vision™ CNN, OpenCV, YOLO, SSD & GANs
Udemy (May 2022)
UC-cea3a356-fa52-46a5-8120-f09bcea73506
Machine learning Deep Learning Model Deployment
Udemy (Oct 2021)
UC-cea3a356-fa52-46a5-8120-f09bcea73506
Applied Artificial Intelligence Club
President
SGA Club, University of South Dakota
University of South Dakota
Masters in Computer Science, AI Specialization (2022 Fall)
Expected Graduation: December 2023
Tribhuvan University
Bachelor in Computer Engineering, KEC (Tribhuvan University) (2014-2018)