Amish Regmi
Software Development Engineer
AWS Marketplace: aws.amazon.com/marketplace
Phone:
940-299-9229
Email:
A Bit About Me
I'm a Software Engineer from Nepal, currently working at AWS Marketplace's Agreements team. I have over 4 years of professional experience building end to end cloud based distributed system solutions in the communication, compliance, transportation, marketplace, and financial sectors of transactions.
​
I'm deeply passionate about the idea of a "Technological Singularity" and am working on personal projects with the umbrella goal of onboarding an AI assistant in order to boost productivity so that I can pursue research into interesting topics such as the nature of consciousness, the potential applications of technological advancements for more safety, security, reducing latency in governmental processes, and gaining further insights into our inner selves.
Work Experience
July. 2022 - Present
Oct. 2021 - July 2022
July 2020 - Oct. 2021
May 2019 - Aug. 2019
May 2018 - May 2020
Aug 2017 - May 2018
Software Development Engineer
AWS Marketplace
https://docs.aws.amazon.com/marketplace-agreements/latest/api-reference/welcome.html
​
Software Development Engineer
Amazon Global Mile
​
Software Developer
IBM Sterling​
Fullstack Developer Intern
IBM Sterling​
Network Assistant
Ramapo College
Research and Marketing Assistant
Alumni Relations
Ramapo College
Projects/ Research
March 2020
Honors Capstone Project
Ramapo College
Story-Share: Developed an android app beta prototype for discovering or sharing different kinds of stories built on an ML approach.
​
How it works:
1. For an uploaded story, utilize ML APIs to summarize the text.
2. Utilize APIs to extract keywords and the emotions from the extracted summary and map the uploaded story to appropriate categories. Utilized some custom labeling here to scale API value from different sources.
3. The user would be able to search stories based on categories and provide reviews.
​
Tools: Django, Android Studio, ML APIs (Rake, ParallelDots), SQL
​
Github: github.com/amishregmi/Story-Share
Sept - Dec 2019
Class Project
Ramapo College
Five-Crowns: A complete console C++ implementation of the game Five-Crowns, followed up with an android implementation including UI elements.
​
How it works:
pages.ramapo.edu/~amruth/teaching/opl/projects/5crowns/problem.html
​
Tools: C++, Android Studio
​
Github:
1. C++ Implementation: github.com/amishregmi/Five-Crowns
2. Android implementation: github.com/amishregmi/Five-Crowns-Android
Oct. 2019
HackRu
Rutgers University
Stock Hub: One stop shop to keep up to date with the stock market behavior by running simulations on past data, tracking latest public social media sentiments, and observing an insightful graphical representation of data.
​
How it works:
It runs stock data simulation based on user inputs of total amount of money to be invested in the selected stock market data and the date, provides latest real time twitter sentiment analysis on the company and a graphical breakdown of the changing sentiments, and a pie chart that effectively lists the different sentiments. The users will be able to easily keep track of the latest trends, and make smarter decisions.
​
Tools: Angular, Cordova, Ionic, Flask, C++, ML APIs (Textblob).
DevPost: devpost.com/software/stock-hub
Feb. - March 2019
Student Web Developer
Ramapo College
Ramapo College Tutor Website: The official tutor finding website for Ramapo College. I delivered the website from scratch.
​
How it works:
- Faculty can login and CRUD tutor details by course.
- Student login and search for tutors by course and schedule.
​
Tools: Express, Node.js, Heroku, Ajax, JQuery.
​
Website: tutoring.ramapo.edu
Fall 2018 - Spring 2019
Research Assistant
Dr. Eman Abdelfattah
Ramapo College
Research Topic: Comparison of performance of machine learning algorithms for a given dataset to understand potential use-case in realtime threat detection systems.
​​
Implementation:
1) Analyzing machine learning algorithm effectives on dataset for intrusion systems in network attacks.
Paper: core.ac.uk/download/pdf/288296483.pdf
2) Analyzing machine learning algorithm effectiveness on DDos attack dataset.
​
Tools: Ipynb, Python, Keras, Random Forest, Support Vector Machine, Logistic Regression etc.
​
Github:
1. github.com/amishregmi/IntrusionDetection
2. github.com/amishregmi/Machine_Learning_CICID2017
​
Research Paper:
Feb. 2019
HoyaHacks
Georgetown University
Wordy: An android app developed to solve a problem I had encountered during my one day layover in Guangzhou. I couldn't understand the signboards and had a lot of trouble with communication.
​​
How it works:
Wordy utilizes google vision api to recognize the text blocks from camera activity, adjusts the custom pixel values to accurately extract the word tapped by the user in the camera. Then, it uses Oxford APIs to extract the word definitions, IBM Watson's translation APIs to translate languages, and Azure's cognitive services for speech to text translations.
​
Tools: Android Studio, Google Vision APIs, Oxford Dictionary APIs, IBM language translation APIs, Azure cognitive speech API for text translation.
​
Github: github.com/amishregmi/Wordy
​
DevPost: devpost.com/software/wordy-2b6rvh
Fall 2018
Class Project
Ramapo College
Geocoding: An effective address book with CRUD functionality and geocoded locations for contacts.
​
How it works:
The project allows the user to add contact information, with their address, through a form. It has a login page and allows authorized users to view all the contacts with their addresses present on a map as a marker, with the tooltip of the marker containing the contact information. The rows of the table containing contact information is clickable and centers the map to the marker of the contact. In the contacts page, you also have the option to Update, or Delete contacts. Delete is implemented as an AJAX request.
Tools: JavaScript, jQuery, Node.js, MongoDb, Mapbox APIs, Jade.
​
Github: github.com/amishregmi/assembler
​​
Project specifications: docs.google.com/document/d/1lfnxI3C_44m9MwZaBxAtfBb9w6jciJvoIsSQw40mXvg/pub
Fall 2018
Class Project
Ramapo College
Assembler and Emulator: Low-level Assembler and emulator for the VC3600 computer.
​​
Project specifications: pages.ramapo.edu/~vmiller/SoftwareDesign/AssemblerNotes.html
​
Tools: C++.
​
Github: github.com/amishregmi/assembler
Nov. 2018
YHacks
Yale University
ML4Connect: A web application platform that leverages ML advancements to offer personalized mapping of clients to therapists.
​​
How it works:
The application saves patient and psychiatrist information on a database using PostgreSQL, and finds the best fit for both, through our own scaling indices, based on text analysis that determines the scale of depression, keywords through TFIDFVectorizer, etc. Should the sentiments analyzed be too low, an email is sent to the emergency contact with the current location of the victim. Also hosts a chat server allowing patients to talk to psychiatrists.
​
Tools: Node.js, JavaScript, jQuery, Python, SQL, google APIs, and some other libraries and frameworks.
​
Github: github.com/anujbastola/ML4Connect
Oct. 2018
HackRu
Rutgers University
BeCalm: A general platform that combines ML capabilities for mental health improvement.
​
How it works:
The inspiration for the project came from the identification of applications of Machine Learning in mental health. The project is designed to help people reduce stress and strive away from depression, by providing personalized recommendations based on their calculated sentiments. Currently, the web application provides 3 options. To enter text and measure the sentiment scores, to take a picture and measure the sentiment scores, or to pick a movie and get movie recommendations through content-based filtering on movies dataset. The scale of the text emotion analysis is from 0-1, the picture emotion detection has 9 emotional categories. The next step is to scale all of the emotions with that of movies and provide recommendations that will help lift your mood up. Additionally, the website will incorporate the function of notifying nearby services, should the emotional level identified be extremely poor.
​
Tools: Python, OpenCV, SQL, HTML, CSS, JavaScript.
​
DevPost: devpost.com/software/calm