Machine Learning Engineer
Fusemachines
Kathmandu, Nepal · In pursuit of wisdom
Sunil Ghimire Linus
Words and data are the two most powerful tools to change the world.
Machine Learning Engineer
NLP
Writer
I build natural language systems by day and write about them at night. The work is below; the writing has a page of its own.
PDF · 69 KB
शब्दहरू सुकेर गए, अर्थ बाँकी छ अझै
आँखा रित्तिएर गए, सपना बाँकी छ अझै
हिँड्दा हिँड्दै बाटो हरायो, पाइला थाकेनन्
साँझ ढल्केर गयो, बिहान बाँकी छ अझै
Sunil Ghimire · read all 5
01 / About
A Human. Being. Noun.
I am an NLP engineer with a soft spot for literature. Most of my day is spent in front of a screen, programming and pulling insight out of large datasets: data analysis, visualisation, and machine learning. I like end-to-end problems: the ones that start with scraping data from an awkward third-party source and end with a business actually understanding its customers.
At night I write. Articles, tutorials, the occasional मुक्तक. Sometimes I build a small personal project just to have something to write about. Sometimes I go to HackerRank and lose an evening to a problem I did not need to solve.
- [email protected]
- Based in
- Kathmandu, Nepal
- Favourite quotation
- Do your best and God will do the rest
- Favourite equation
- x² − x + 41
02 / Focus
What I work with
Grouped by what I actually reach for, not by a percentage bar.
Languages
Operating systems
Version control
ML / DL
Computer vision
Visualisation
NLP
MLOps
Web
Databases
Vector databases
Graph databases
Development
Documentation
Beyond the code
03 / Experience
Where I have been
Graduate Teaching Assistant, AI & ML
Herald College Kathmandu
04 / Publications
Peer reviewed, and one book
Work published with co-authors at Fusemachines, the Institute of Engineering at Pulchowk, and elsewhere.
-
Conference paper · 2024
A Comparative Study of Transfer Learning Approaches for Strengthening Face Antispoofing Security
ACMLC 2024, the 6th Asia Conference on Machine Learning and Computing, Bangkok. ACM Digital Library.
Read the paper -
Conference paper · 2024
Application of Nepali Large Language Models to Improve Sentiment Analysis
ICCMB 2024, the 7th International Conference on Computers in Management and Business, Singapore. Pages 144 to 150. A benchmark dataset for Nepali sentiment analysis, reaching an F-score of 0.88.
Read the paper -
Journal paper · 2023
SMS Spam Detection using Relevance Vector Machine
Procedia Computer Science, volume 230, pages 337 to 346. Elsevier, open access. A relevance vector machine reached an F1 score of 0.975 on the task.
Read the paper -
Book · 23 May 2020
Ultimate Guide to Python Basics
Amazon Kindle edition. Python lets you write programs in fewer lines than most languages, and this is where I explain why that matters.
View on Amazon
05 / Services
How I can help
Four things I take on outside of full-time work.
01
Data science
Machine learning pipelines and data products built to answer a real question about your business or your customers.
02
Technical writing
Journal articles, technical manuals, training guides, and assessments, written to be read rather than skimmed past.
03
Course design
Online courses planned, sequenced, and paced so learners finish them instead of abandoning them at lesson three.
04
Talks and mentoring
Conversations about where NLP is heading, and guidance for people trying to get into the field.
06 / Selected work
Projects and writing
A mix of shipped code, from-scratch implementations, and things I wrote up afterwards.
Project
Resume Parser
Scans a CV, analyses it, and pulls out the details recruiters actually need.
View repo
Project
Object Tracker with YOLO
Real-time object tracking built on YOLO, for video processing and computer vision work.
View repo
Project
Blood Cancer Detection
Detecting leukemia, lymphoma and myeloma from blood imagery. Reached 91% validation accuracy.
View repo
Project
Credit Card Fraud Detection
Separating fraudulent from genuine transactions using logistic regression, random forests and an autoencoder.
View repo
Project
Three Real-World Data Science Projects
Predicting airline prices, password strength, and stock prices: three end-to-end builds in one repo.
View repo
Project
Vehicle Detection and Counting
A traffic management tool that detects and counts vehicles from video for intelligent transport systems.
Read write-up
Project
Car and Plane Detection
A convolutional neural network that tells cars from planes at 87% validation accuracy.
View repo
Project
Face Recognition
Face detection and recognition built with Python, NumPy and OpenCV.
View repo
Project
Movie Review Analysis
Sentiment classification trained on 25,000 reviews and tested on 25,000 more.
View repo
Project
Movie Recommendation
A KNN recommender over movie titles and ratings, using distance between items.
View repo
Project
Auto Tweet Bot
A Selenium automation that tweets a transcript line by line, on its own.
View repo
Project
Air Route Tracker
An offline tool for identifying airline paths, simple enough for anyone to use.
Read write-up
Project
Tic Tac Toe
Built on minimax, the same algorithm that turns up in AI, finance and game theory.
Read write-up
Project
Hangman
A word-guessing game for two or more players, written up as a step-by-step guide.
Read write-up
Writing
PCA from Scratch
Finding the direction of maximal variation by diagonalising the covariance matrix and minimising reconstruction error.
Read article
Writing
KNN from Scratch
Implementing k-Nearest Neighbors in Python without reaching for a library.
Read article
Writing
50 Questions on Statistics and ML
Fifty tricky questions I collected and answered, to sharpen the fundamentals.
Read article
Writing
An Intelligent Autopilot System
On autonomous steering, a corner of self-driving research that stays largely unexplored.
Read article
Writing
Linear Regression from Scratch
A supervised method for finding the relationship between continuous variables, built up from nothing.
Read article07 / Contact
Say hello.
Let's not be strangers. Questions about a project, a job, or something I wrote are all welcome. I read everything and reply as soon as I can.
- Email[email protected]
- Phone+977 984 107 0311
- AddressTokha Municipality-09, Kathmandu