Education
Carnegie Mellon University
Aug 2022 — May 2024
Focus on deep learning, NLP, and AI systems development.
Addis Ababa University
Sep 2016 — May 2020
Strong foundation in software engineering, algorithms, and system design.
Recognized as one of the top 3% of freelance developers on Toptal's exclusive talent network.
Certifications
Hugging Face
March 2026
Certificate of Excellence for successfully completing the Hugging Face Deep Reinforcement Learning Course.
LangChain Academy
January 2026
Certificate in building deep agents with LangChain.
LangChain Academy
January 2026
Certificate in LangChain foundations for building applications with LLMs using Python.
LangChain Academy
October 2025
Certificate in LangGraph foundations for building stateful, multi-actor applications with LLMs.
Experience
Software Engineer II, AI at Motive
Mar 2025 – Present
- Building and maintaining a multi-tenant car dealership platform used by 400+ dealerships across the U.S. and Canada, serving millions of users, leveraging Python (LangGraph, FastAPI, FastMCP), Ruby on Rails, and Next.js.
- Developed an agentic conversational system using LangChain, LangGraph, LangSmith, FastAPI, and Pydantic, capable of creating and updating webpages, performing content analytics, and generating SEO-optimized blog posts, enabling dealership admins to scale content automation.
- Built a Vehicle Trade-in Evaluator by training a deep feedforward neural network on 2 million vehicle inventory records in PyTorch, encoding vehicle attributes with Sentence Transformers (all-MiniLM-L6-v2 embeddings), achieving 9.02% MAPE on trade-in value prediction.
- Fine-tuned lightweight reranker models using PyTorch, Sentence Transformers, Hugging Face, and Vertex Workbench to improve search result relevance and boost conversion rates.
- Designed and deployed MCP servers for dealerships to utilize our platform tools and integrate them with Claude, ChatGPT, and Cursor AI assistants.
- Set up training and deployment pipelines on GCP Vertex AI (Cloud Storage, Artifact Registry, Vertex Training, Vertex Model, Vertex Endpoint) and used Weights & Biases for experiment tracking and monitoring.
- Utilized the pytest framework to test AI microservice functionality, structuring tests with the Arrange–Act–Assert pattern for clarity and maintainability.
Software Engineer, AI at eezly
Aug 2024 – March 2025
- Worked on eezly, a grocery price comparison application used by over 30,000+ users, leveraging ASP.NET Core Web API, Python (FastAPI), PyTorch, and cloud-based microservices to build scalable, AI-driven app.
- Built a Recipe Recommendation System using LangChain, OpenAI, and the Recipe1M+ dataset, creating a Retrieval-Augmented Generation (RAG) system to suggest recipes based on the products users purchase. Incorporated the Weaviate vector database to enhance search and recommendation.
- Employed PyTorch and Hugging Face to train hierarchical machine-learning models for classifying retail products from various stores (e.g., Walmart) into aisles, categories, and subcategories.
- Integrated Gorse, a recommender system, and contributed to open-source recommender systems.
- Designed and implemented RESTful APIs for inventory management using n-tier architecture and developed a single-page application with React.js.
- Implemented OAuth 2.0 client-credential flow using OpenIddict for secure machine-to-machine communication, Single Sign-On (Firebase, Cognito), and ASP.NET Core Identity for user management.
Teaching Assistant at Carnegie Mellon University
Jun 2024 – Dec 2024
- Held weekly office hours to assist students taking 11-785 Introduction to Deep Learning, a PhD-level course, with deep learning concepts, homework, and coding challenges in PyTorch.
- Reviewed and improved homework assignments that help students build RNNs and GRUs from scratch using a custom reverse-mode automatic differentiation framework to register computations and backpropagate.
- Worked with other TAs to run ablations using Weights & Biases (wandb) to track and analyze the performances of models students should try out.
- Configured AutoLab, a testing platform, and assisted in releasing homework assignments to evaluate student submissions.
Research Assistant in Machine Learning at Empathic Computing Lab
May 2021 – Jan 2022 · Auckland, New Zealand
- Empathic Computing Laboratory (ECL) is an academic research laboratory directed by Prof. Mark Billinghurst at the University of South Australia in Adelaide, Australia, and the University of Auckland in Auckland, New Zealand.
- Collaborated with Ph.D. students to refine methods for detecting emotions from physiological signals.
- Conducted extensive literature reviews and analyzed the performance of various machine learning and deep learning models, applying rigorous hyperparameter tuning. Authored a 14-page paper (IUI - ACM).
LeetCode
LeetCode progress — algorithms and data structures: