🏋️ Binyam Sisay Atnafu.
Addis Ababa, Ethiopia.
Computer Engineer, BSc
Austin, Texas, United States
Email: binasisayet8790@gmail.com
Phone Number: +13466209880
Linkedln: https://linkedin.com/in/bina-man
Python
SQL | NoSQL(Mongo DB)
AWS
Github | DVC | Heroku
Git
Tensor Flow,
OpenAI
OpenCV
React
Flutter
Pythorch
About me
Over the years, I've dived deep into the realms of Artificial Intelligence (AI) and machine learning. At Adludio, I tackled huge datasets, refining and enhancing machine learning models that were crucial for various research applications. My stint at the Ethiopian Public Health Institute was especially eye-opening. There, I applied advanced machine learning techniques to study important health topics, adding depth and innovation to traditional research methods. Another significant chapter in my journey was with ICOG-Labs. I was entrusted with the responsibility of redeveloping a machine learning product. This product was designed to identify images, and I improved it using the latest AI techniques. Currently, my passion has me delving into the intriguing world of generative AI. This cutting-edge technology is shaping the future of AI, and I'm deeply invested in exploring its vast potential. Through all these experiences, my dedication to harnessing the power of AI for practical solutions has only grown stronger.
Work Experience
- Mid-Level Data Engineer ((Apr 2022 - Current):
Adludio - Contract
At Adludio, I played a pivotal role in harnessing the power of big data, processing over 10 terabytes of historical information to fuel various machine-learning models and analytics. Delving into the vast digital landscape, I spearheaded the scraping of content from over 20 million sites/publishers, a move that greatly enriched and amplified the potency of our historical data. My expertise in AWS was exemplified as I developed intricate data marts utilizing AWS Athena, Glue-crawler, Spark, and other AWS managed services. This facilitated the construction of robust warehouses, databases, and data lakes on platforms like AWS S3 Bucket, RedShift, and Snowflake, ensuring rapid and reliable data accessibility. Experimenting with data models, such as Kimball, Inmon, and Datavault architectures, I yielded an agile and resilient data model tailored for the company's operations. Additionally, I pioneered the deployment of diverse pipelines for projects encompassing analytics, machine learning, and other ETL processes. My contributions also extended to the development of analytics and reporting dashboards, leveraging React, FastAPI, and Postgres.
My architectural prowess was evident as I orchestrated an end-to-end AWS data pipeline and data warehouse, integrating components like Glue, Athena, Spark, S3, Lambda functions, and EMR clusters. My engineering initiatives included crafting a data warehouse with Hive, managing Hive tables, and devising Spark code in Python for expedited data testing and processing. By modeling, aggregating, and triangulating multiple log datasets, I ensured they were seamlessly populated to RDS databases and established a dependable data warehouse on both AWS and Snowflake. These strategic projects not only streamlined various pricing models but also significantly enhanced the efficacy of campaigns. This optimization directly bolstered key performance indicators, including cost-per-engagement and cost-per-impression, culminating in substantial financial savings for the company by better targeting effective targets.
Key points:
Managed over 10 terabytes of data, enhancing machine learning capabilities.
Scraped content from millions of sites to enrich historical data.
Deployed AWS tools for robust data storage and retrieval systems.
Pioneered data structures tailored to the company using different models.
Established efficient pipelines for analytics and machine learning.
Developed reporting tools with React, FastAPI, and Postgres.
Orchestrated a comprehensive AWS data pipeline and data warehouse.
Optimized pricing models, enhancing campaign effectiveness and financial savings.
- Machine Learning Engineer (Dec 2021 - Apr 2022):
Ethiopian Public Health Institute (EPHI) - Paid Internship
At the Ethiopian Public Health Institute, I took on the challenge of harnessing advanced Machine Learning (ML) algorithms to drive transformative changes in Demography and Epidemiology research. My role involved the integration and analysis of data from the Ethiopian Demographic and Health Survey across multiple years, focusing on the complexities of under-five mortality rates. Leveraging the power of three distinct ML models, I discovered that Gradient Boosting (XGBoost) stood out in its predictive accuracy, offering a modern alternative to conventional Logistic Regression methodologies.
Diving deeper into the data, I identified a myriad of influential factors affecting Under-five mortality, spanning individual attributes like breastfeeding practices to broader environmental metrics such as aridity and vegetation indices. Recognizing the nuances of data interpretation, I ensured that correlations were meticulously vetted for causality. To achieve this, I employed the innovative NOTEARS algorithm, which facilitated the extraction of clear causal relationships among diverse features.
These analyses revealed critical intervention points. For instance, community dynamics, such as the number of children in households and water sanitation practices, combined with individual factors like birth intervals, emerged as direct contributors to under-five mortalities. Drawing on these insights, I championed a holistic, evidence-driven approach to health interventions. I proposed and advocated for a multi-pronged strategy that emphasized nutrition, family planning, and infrastructure improvements in water and sanitation. My contributions not only set a new standard for data-informed health strategies but also laid the groundwork for sustainable, impactful health initiatives aimed at reducing under-five mortality rates in Ethiopia.
Key points:
Utilized advanced Machine Learning to innovate in Demography and Epidemiology research.
Analyzed multi-year data to focus on under-five mortality rates, endorsing XGBoost for its accuracy.
Employed the NOTEARS algorithm to extract clear causal relationships in data.
Identified community and individual factors as pivotal contributors to under-five mortalities.
Advocated for a comprehensive, evidence-based approach to health interventions.
Emphasized the importance of nutrition, family planning, and water sanitation in health strategies.
Set a precedent for data-driven health strategies in Ethiopia.
Laid foundational groundwork for sustainable health initiatives targeting under-five mortality reduction.
- Software developer (Mar 2021 - Oct 2021):
At CNET, I seamlessly blended technological innovation with user-centric designs, driving the company's mobile application portfolio to new heights. My relentless pursuit of excellence, adaptability, and an innate aptitude for debugging enabled me to consistently deliver solutions that not only resonated with our users but also drove significant business value. By harnessing the power of Dart and Flutter, I was at the forefront of pioneering CNET's transition into cutting-edge mobile solutions, ensuring the apps remained relevant, efficient, and impactful in a rapidly evolving digital landscape.
Key contributions include:
Orchestrating the integration of diverse payment gateways, streamlining e-commerce functionalities.
Pioneering the integration of popular social services, enhancing user engagement and connectivity.
Implementing real-time push notifications, elevating user interaction and app retention rates.
Leveraging Flutter's cross-platform capabilities, ensuring a consistent and high-quality user experience across all devices.
Exemplary debugging skills, ensuring seamless app performance and swift resolution of any issues.
- Software-developer (Aug, 2019 - Mar, 2021):
Addis Zeybe - Intern/ Part-time
At Addis Zeybe, I started as a Web Developer and built the company's main website. I added tools to help the site show up in search results and used Google Analytics to see how people used our site. Later, I moved to a Software Tester role where I checked another website to make sure it worked well and was easy for users.
Key points:
Built and set up Addis Zeybe's main website.
Added tools to improve the website's search ranking.
Used Google Analytics to track website use.
Tested another website to ensure it worked properly.
- Computer Engineer (Mar 2018- July 2018):
ICOG-Labs - Intern
At ICOG-Labs, I embarked on an enriching internship journey as a Computer Engineer. During my tenure, I took on the challenge of revamping the company's official website, ensuring it reflected the latest updates and resonated with the brand's identity. Venturing further into the realms of Machine Learning, I had the privilege to overhaul one of their flagship products focused on image classification. I employed both simple linear regression and the more advanced convolutional neural network (CNN) methodologies for image classification, enhancing the product's efficiency and accuracy.
Key points:
Successfully updated and modernized ICOG-Labs' official website.
Redesigned a pivotal machine learning product specializing in image classification.
Utilized simple linear regression for foundational image classification.
Integrated convolutional neural network (CNN) techniques to advance the product's classification capabilities..
Summary
I have been working on developing software and currently am looking for carriers in the fields of Machine Learning and Data Engineering. I started my way
to software development by Node.js as a backend developer, I used EJS as a dynamic rendering tool for the front end and mongo as a database,
I have built an e-commerce site using node.js, which currently is not hosted on a server. Making my way to software I joined Addis Zeybe news and Media
company last year in august and maintained their website which was made by WordPress. They then changed the website to be built by typescript and
javascript, where I was assigned as a software tester and currently a webmaster @Addis Zeybe. I then joined CNET as a Researcher and Software developer
and currently working there full time.
Education
Data Science and Machine Learning,
Graduated from 10Academy in Machine Learning Engineering. In the intensive 12 weeks of training, I was able to work on 9 real-world projects. I was able to accomplish 2 of the projects with a group of four people and the remaining 7 projects as an individual task. The projects can be found here. The training was for 12 weeks of an intensive project. The projects were designed in a way the new industry requests and I were able to finish up real-world projects.
BSc. (Electrical And Computer Engineering)
At Addis Ababa University, a premier institution, I successfully completed my Bachelor's degree in Electrical and Computer Engineering with Great Honor, securing an impressive CGPA of 3.7/4. This achievement not only underscores my commitment and hard work but also showcases my strong technical aptitude. My academic path was shaped by a profound engagement with subjects that now form the bedrock of my expertise. Mathematics, spanning from Calculus I, II, and III, to Advanced Probability, was an area of particular fascination for me. My technological pursuits extended to areas like Digital Signal Processing, Operating Systems, Database Management, Software Engineering, and Computer Security, each of which deepened my understanding of the digital realm. My programming skills were honed through comprehensive courses in languages like C, C#, C++, and Java. Furthermore, I delved into specialized subjects like Object-Oriented Programming, Data Structure, Algorithms, Robotics, and Computer Security, each playing a pivotal role in shaping my multifaceted approach to tech innovation.
Stanford university
The skills I gained in this course were Logistic regression, Artificial Neural Networks, Machine Learning Algorithms, and Machine learning.
University of MICHIGAN
The skills I gained in this course were Python Programing, Python syntax, and semantics, Computer Programming.
Projects
Cloud-Big data processing
Cloud architecture and orchestration designs and implementation on AWS.
Under five Child Mortality
This project aims to forecast and see which features are more contributing to child mortality in Ethiopia based on high dimensional EDHS data. Applied multiple modeling and see the results accordingly. Triangulated 5 years of EDHS data and applied year-long analysis and see trends. Applied PCA and Autoencoder to see the linearity among the variables.
The key to success for telecommunication companies is to segment their market and target the content according to each group. This golden rule is relevant to the various areas of business. Speaking about telecommunication, there are four segmentation schemes of primary importance: customer value segmentation, customer behavior segmentation, customer lifecycle segmentation, and customer migration segmentation.
This project was made in a team of 10 people me included. Initially starting with data preprocessing, we used modeling using Tensorflow Even though there are two basic ways of building for such kind of data, our team decided to use CNN and bi-directional RNN to transcribe the voice. Meta-data generation was also made but as a result of the size of Tensorflow it wasn't hosted.
The key to success for telecommunication companies is to segment their market and target the content according to each group. This golden rule is relevant to the various areas of business. Speaking about telecommunication, there are four segmentation schemes of primary importance: customer value segmentation, customer behavior segmentation, customer lifecycle segmentation, and customer migration segmentation.
Using A/B testing to test if the ads that the advertising company ran resulted in a significant lift in brand awareness.Comparing machine learning models vs A/B testing gave me insights on what to use in which particular problem.
Precise sales prediction is an essential and inexpensive way for each company to augment their profits, decrease their costs, and achieve greater flexibility to changes. In other words, exact sales forecasting is utilized for capturing the tradeoff between customer demand satisfaction and inventory costs. Especially, for the pharmaceutical industry, successful sales forecasting systems can be very beneficial, due to the short shelf-life of many pharmaceutical products and the importance of product quality which is closely related to human health.
Mobile application
Mobile application. made using Flutter. The application uses flutter for building the UI, and also uses .net as a backend. SQL for database and uses firebase for OTP. The aim of the app is to build an eCommerce app for a huge local company called CNET. check the app