Machine Learning
Protein Structure Prediction using Deep Learning — (Mar,2023 - Nov,2023)
https://github.com/usman61/protein-structure-prediction- Objectives: To Design and implement a RNN based model that accurately predicts protein secondary structures.
- Started by exploratory data analysis, preprocessing the dataset, ensuring its quality and suitability for the Deep learning model.
- Finally, Designed and trained an GRU based model with CNN, which is a neural network and assessed the performance of the model by evaluating its accuracy 74.77%.
Anomaly Detection using Autoencoder — Supervised ML (March, 2023 )
Dataset: https://www.kaggle.com/datasets/mlg-ulb/creditcardfraud/- Developed an anomaly detection system using an autoencoder.
- Objective of the project was to identify fraudulent transaction made by credit card.
- Started by exploratory data analysis, preprocessing the dataset, ensuring its quality and suitability for the autoencoder model.
- Next, designed and trained an autoencoder, which is a neural network architecture capable of learning efficient representations of the input data.
- Finally, I assessed the performance of the anomaly detection system by evaluating its accuracy (99%) and precision(87%).
Salary Range Prediction — Supervised ML (May, 2022 )
- Develop the model to predict the individuals earning above 50k or below based on various input features using Naive Bayes.
- Explored and analyzed the dataset to gain insights into the relationships between the features and the target variable (salary range).
- Techniques such as Feature Engineering, one-hot encoding and feature scaling were applied as appropriate.
Skin Segmentation using Decision Tree — Supervised ML (April, 2022 )
Dataset: https://archive.ics.uci.edu/dataset/229/skin+segmentation/- Objective: To accurately distinguish between skin color and non skin color.
- Implemented a skin color model using a Decision Tree algorithm.
- Evaluated the performance of the model using appropriate evaluation metrics (e.g., precision, recall, F1-score)
Data Analytics
Dashboard for Sky Transport — Power BI (July, 2023 )
https://github.com/usman61/PowerBI- Developed a Dashboard to show top level brief overview of Sky Transport Data with several set of visualizations.
- KPIs for Total Miles, Revenue, Cost
- Revenue Vs Miles by Shipping State
- Revenue Vs Miles by Shipping City
- Total Trips Vs Trips by Trip Types
- % of Trips by Shipping State
- Revenue Miles vs Total Miles
Data Analysis on Coronavirus Cases — Power BI (June, 2023)
https://www.worldometers.info/coronavirus/- Connected Power BI to a prominent COVID-19 website for real-time tracking of cases.
- Leveraging my expertise in data integration and visualization.
- Used Power Query Editor to transform and preprocess the data.
- Generated different charts of COVID-19 patients (death cases,recovered cases) country wise and continent wise.
Airline On-Time Statistics and Delay Causes — Big Data Analytics (May, 2022)
- Applied different statistics on airline data (collected from Bureau of transportation statistics, US)
- Used Hadoop (Map Reduce) to realize the various concerns related to airlines which includes:
- Reasons behind delay causes in flights, to predict flight cancellation ratio, busiest airport, airline and several others.
- All these operations were performed on Google Cloud Platform by configuring an instance of Linux Virtual Machine and Hadoop Distributed File System.
Natural Language Processing (NLP)
Compute Distance Among various Pashto Dialects — NLP (Dec,2022)
- IPA transcriptions of 24 words are provided from various Pashto dialects of 11 regions/cities along with their comparative words in Urdu/English.
- Developed the code (didn’t use ready-made libraries) for computing Levenshtein Distance (Edit distance).
- Used python and jupyter notebook to develop the algorithm.
Web Applications
Weighbridge Automation System — Veevo Tech (Sep,2020 – Oct,2020)
https://ssw.veevotech.com/- Developed Web Portal for Sindh Solid Waste Management Board (SSWMB) Govt. of Sindh, Pakistan entity.
- To keep and anlyze the record of garbage weight which trucks dump at their landfill sites.
- Portal provides reporting (Site wise, Customer wise,Vehicle Wise) in Excel and Pdf formats.
PIPS — Veevo Tech (Feb,2020 – March,2020)
https://pips.veevotech.com/- Developed Web Portal and API of Mobile application for Pakistan Institute for Parliamentary Services (PIPS).
- To evaluate the applications submitted by Parliamentarians (Members and Functionaries) from a mobile application to the corresponding (PIPS) department in the Web Portal.
- Notified all the Members of the corresponding (PIPS) department via Email when an application is submitted by Parliamentarians.
- Chat Box to interact with Parliamentarians regarding their submitted application.
- To register and manage PIPS Employee by Admin.
Notes Pool — Veevo Tech (April,2020 – May,2020)
https://emp.veevotech.com//Our objective was to design a digital notebook in which users could easily create a notebook which contains multiple notes, attach documents, images and share their notes/notebook with their colleagues within an organization or make their notes public as well if required. Configured Editor js a block-style editor to create notes.
Dakia — Veevo Tech (June,2020)
In this project we have developed an email sending service through code using PHP Mailer library. It’s basically an open API which developers utilize to send email to anyone. Developers have to just push (email subject email body and the time at which they want to send email to the recipient) to this API. We have also defined some predefined html templates (i.e. forgot password, email verification, invoice generation) and identified few place holders to replace it with developer provided text whenever to send an email to a specific recipient while using predefined template.