Iman Asfaw
345 Murray Farm Rd, Fairview,TX
Summary
Experienced Senior Software Engineer, well-versed in OOP concepts and design patterns. AI/ML and Software Engineer with over 7 years of experience building scalable microservices and cloud-native applications. Specialized in developing robust backend systems and integrating AI/ML, GenAI, and LLM models into production. Proficient in Python, AWS, and modern MLOps, with a strong focus on reliability, automation, and turning complex data into intelligent, user-facing solutions that drive real business impact.
Overview
9
9
years of professional experience
1
1
Certification
Work History
Senior Software Engineer
CapitalOne
05.2025 - Current
- Trained and deployed a Random Forest model to classify CPO documents, improving automation from 81% to 94% accuracy in the auto-loan workflow.
- Built a post-automation notes feature to enhance agent visibility and auditability of loan applications.
- Implemented BDD live dependency tests to prevent regressions and ensure new changes did not break business requirements.
- Contributed to cybersecurity and production support by mitigating vulnerabilities and monitoring live applications to reduce operational risk.
- Integrated the Notes API into the POI Orchestrator to mark successful task autocompletion, significantly improving auditing visibility and operational transparency.
- Implemented live dependency testing in the QA pipeline to validate new merges against real service dependencies, enabling early detection of a production-critical bug that I documented, scoped, and resolved.
- Fixed structural test logic that previously caused premature application exits, ensuring correct enforcement of business rules and improving overall system reliability.
- Led an urgent ETB secret-rotation initiative by classifying, rotating, and validating secrets across all services, meeting strict compliance and timeline requirements.
- Enhanced VVAL automation by enabling Notes integration to track and audit successful task execution end-to-end.
- Authored and maintained Confluence documentation for all service secrets, including ownership and rotation history, improving long-term operational clarity and knowledge sharing.
- Strengthened security posture by updating Docker images, upgrading vulnerable libraries, remediating pydf issues in the Document Classifier service, and fixing brace-expansion vulnerabilities in the Document Viewer.
- Delivered key business enhancements, including implementing adjustmentWholesaleValue to align pricing with mileage-based depreciation and adding commercial-vehicle checks to enforce lending policy.
- Contributed to Capital One’s community initiatives by volunteering at a local homeless shelter, supporting corporate social responsibility efforts.
Software Engineer
Mereb technologies
02.2022 - 09.2023
- Designed a feature for the cryptocurrency module that allows users to view wallets and transfer credit using Express.js and Node.js, helping users manage their crypto accounts more effectively.
- Developed a solution for the admin module to efficiently retrieve user data from a MongoDB database using JavaScript and Node.js, resulting in an 80% improvement in performance compared to the previous version.
- Dockerized backend server with MySQL and Redis integration, streamlining testing and increasing development speed by 70%.
- Contributed to all aspects of Software Development life cycle resulting in efficient development, integration testing and deployment of code which resulted in promotion and client satisfaction.
Machine learning Engineer
iCogLabs
08.2016 - 02.2022
- Contributed to all Machine learning life cycle of train, test and deploy of AI/ML applications resulting in high client satisfaction.
- Led the end-to-end machine learning lifecycle of multiple AI/ML applications, covering data processing, model development, testing, and deployment, ensuring high client satisfaction through robust solutions.
- Streamlined model deployment with extensive MLOps practices, including automation of model retraining, pipeline monitoring, and deployment orchestration using tools like Docker and Google AI Platform.
- Engineered scalable, cloud-based infrastructure for model deployment and monitoring, reducing deployment time by 25% and enabling automated performance tracking and alerting for model drift.
- Processed tabular datasets using R and Python for the child mortality prediction model, enhancing prediction accuracy.
- Trained Neural Networks and Random Forest models to predict child mortality rates and Vitamin A coverage with R2-squares of 0.98 and 0.96, respectively, and performed data analysis and feature extraction to provide critical insights for field experts.
- Conducted prediction interpretation using LIME, helping field experts understand and interpret the model's predictions.
- Developed a conversational chatbot for our business domain by integrating UI with Rasa chat engine working on GPT which helped our client drastically cut the cost of traditional call centers.
- Developed and deployed Wav2letter model on Google AI Platform, resulting in a 30% increase in prediction accuracy and a 25% reduction in deployment time for various applications, which enhanced overall system performance and scalability.
- Built a system that extracted important information from scientific papers using Python, tesseract and BERT to help our research employees read scientific papers easily which resulted in performance improvement.
- Contributed to developing a highly efficient AI algorithm in C++ for MOSES, achieving 98% accuracy.
- Developed a Scheme-based database querying module which involves graph learning for AtomSpace,a Graph database, increasing performance by nearly 70%.
- Built a fake news classifier using Idris and Haskell with 95% accuracy, demonstrating the languages' effectiveness in machine learning.
- Tested the Opencog AGI agent with a Ping-pong OpenAI gym environment using Python, helping iCogLabs secure significant funding through a client demo.
- Conducted unit and system testing with CxxTest, and performed code fixes and enhancements for future releases and patches.
Fatima ML Research Fellow
Baylor College of Medicine
03.2021 - 11.2021
- Developed an unsupervised AI algorithm in Python for genetics research, identifying differences in human cell datasets to distinguish between cancer-affected and non-affected cells, earning a $1500 award.
- The model used Anomaly detection to achieve the result.
- Discovered key cancer-related genetic components using an AI model on real cell datasets, achieving 90% accuracy in a published research paper accepted at workshops.
- Conducted A/B testing to evaluate and compare machine learning models, leading to the detection of genetic regions responsible for colon cancer by identifying the most effective algorithm.
Education
Master’s - Computer Science
Maharishi International University
07.2026
Bachelor’s - Software Engineering
Addis Ababa University
02.2020
Thesis - Sequence to Sequence models for Amharic Speech Recognition
Addis Ababa Institute of Technology
12.2019
Skills
- Python
- Java
- Node
- MongoDB
- SQL
- OpenCV
- CI/CD
- Docker
- Kafka
- Tensorflow
- R
- DVC
- Jupyter
- MLflow
- MapReduce
- Keras
- Numpy
- Pandas
- Linux
- REST API
- Data Structures and Algorithms
- OOP
- OpenAI Gym
- AWS
- IAM
- EC2
- Lambda
- SQS
- SNS
- S3
- SageMaker
- Bash
- Spring boot
- Spacy
- HuggingFace
- NLTK
- Hadoop
- Spark
- Rapid Miner
- Github Actions
- Xgboost
- Classification
- LLM
- BERT
- Regression
- Clustering
- Time series
- Hypothesis testing
- Cross validation
- Feature selection
- Feature extraction
- Summarization
- Explainability AI
- GPU
- CUDA
- ETL
- Swagger
- Functional Programming
Certification
- Sequence Models by deeplearning.ai, 98%, https://coursera.org/share/4b3f38498c2f6e6928a9741031791a15
- Convolutional Neural Networks by deeplearning.ai, 98%, https://coursera.org/share/3f18a21ad25c74ba94e1a52ff0effaf9
- Improving Deep Neural Networks by deeplearning.ai, 98%, https://coursera.org/share/a1f17a091ffa75771a877ebd3c1b3f02
- Neural Networks and Deep learning by deeplearning.ai, 100%, https://coursera.org/share/afca5f3624ba2cd63a0b53357a595901
Awards
- Black Air Summer Research Grant, Selected for a research grant for my project 'Unsupervised annotation of differences between genomic dataset using deep Neural Networks.' Received $1500 as a grant.
- Fatima Al Fahri Predoctoral fellowship, Selected to become a fellow for pre-doctoral research.
- Inclusion@RSS fellowship, Selected for Robotics Science and system fellowship.
- OurCS 2019 at Carnegie Mellon University, Selected for a week-long research workshop at CMU Pittsburgh. Advised by Professor Norman Sadeh, Abhilasha Ravichander, Aerin Zhang. Topic: AI for Privacy.
- IEEE VIS diversity and inclusion scholarship
Research
Addis Ababa Institute of Technology, 09/01/18, 12/01/19, Sequence to Sequence models for Amharic Speech Recognition, Developed an Amharic Speech recognition model to convert Amharic audio into its corresponding text. Presented this work at the Black in AI 2019 workshop co-hosted with NeurIPS 2019.
Timeline
Senior Software Engineer
CapitalOne
05.2025 - Current
Software Engineer
Mereb technologies
02.2022 - 09.2023
Fatima ML Research Fellow
Baylor College of Medicine
03.2021 - 11.2021
Machine learning Engineer
iCogLabs
08.2016 - 02.2022
Master’s - Computer Science
Maharishi International University
Bachelor’s - Software Engineering
Addis Ababa University
Thesis - Sequence to Sequence models for Amharic Speech Recognition
Addis Ababa Institute of Technology