AppliedAIEngineer&MachineLearningResearcher
I design and deploy production-ready AI systems powered by LLMs, Retrieval-Augmented Generation (RAG), and agentic workflows. From research to real-world applications, I transform emerging AI capabilities into reliable products that solve meaningful problems.
FromAIResearchtoProductionSystems
I'm a Software Engineering graduate from Addis Ababa Science and Technology University with experience spanning machine learning research, applied AI engineering, and scalable backend development.
At iCog Labs, I researched biologically inspired alternatives to backpropagation for sequence modeling and large language models. Alongside research, I've built and deployed AI-powered systems across education, travel, developer productivity, and business automation.
My work focuses on designing AI systems that move beyond prototypes—systems that are reliable, measurable, and capable of delivering value in production environments.
I am particularly interested in:
BringingAIIdeastoProduction
Applied AI Development
LLM-powered applications designed for real-world use. From conversational agents to intelligent automation, I build AI that ships.
Retrieval-Augmented Generation
Knowledge systems, semantic search, and domain-specific assistants that ground LLM outputs in your data for accuracy and relevance.
Agentic Automation
Multi-step AI workflows using n8n and modern orchestration tools. Automating complex business processes with intelligent agents.
Backend Engineering
Reliable APIs and infrastructure supporting intelligent systems. FastAPI, Node.js, PostgreSQL, and cloud-native patterns.
AI Research
Exploring learning algorithms, evaluation strategies, and resilient AI systems. Bridging the gap between research advances and production reality.
WhereI'veWorked
Machine Learning Researcher Intern
iCog Labs
- Researched Predictive Coding architectures as alternatives to backpropagation for LLMs.
- Designed energy-based objective functions and latent update mechanisms.
- Built modular training and evaluation pipelines.
- Evaluated models using BLEU, BERTScore, and custom metrics.
- Tuned models using Optuna.
Applied AI Engineer
Independent Consultant
- Designed and deployed LLM-powered systems using RAG, embeddings, and agentic workflows.
- Built AI educational systems enabling question generation and automated scoring.
- Developed domain-specific RAG assistants.
- Integrated LLM APIs, vector databases, and automation tools into production environments.
Backend Engineer
Dynamic Technologies
- Reduced latency by approximately 80% by migrating a gaming platform from external APIs to in-house data pipelines.
- Designed optimized schemas and indexing strategies.
- Built microservices supporting personalized learning experiences.
Technical Lead
GDG On Campus AASTU
- Led 70+ developers and coordinated technical mentoring initiatives.
- Organized project reviews and engineering evaluations.
- Facilitated sessions on AI and software engineering practices.
AISystemsI'veBuilt
Codebase Intelligence Q&A
RAG-Powered Developer Assistant
Built a production-ready RAG system capable of understanding entire codebases and answering developer questions conversationally.
- Indexed 40,000+ files in under 4 minutes.
- Reduced hallucinations by ~60%.
- Combined BM25 and embeddings for hybrid retrieval.
- Implemented LLM-based reranking.
- Supported multi-turn conversations.
AI Educational Platform
Arabic Learning Assistant
AI-powered educational system that generates questions from curriculum materials and provides personalized feedback.
- Automated question generation.
- AI scoring.
- Parent-generated assessments.
- Personalized recommendations.
- Parent reporting dashboards.
Voice Travel Planner
Agentic Travel Assistant
A conversational travel planner that transforms natural dialogue into complete itineraries.
- Voice interactions.
- Flight and hotel recommendations.
- Personalized activity suggestions.
- Automated itinerary emails.
Real Estate RAG Assistant
Domain-Specific Knowledge System
Semantic property retrieval and context-aware responses for real estate domain.
- Semantic property retrieval.
- Context-aware responses.
- Domain-constrained generation.
Workflow Recovery Research
Safe Autonomous Failure Detection and Recovery
Investigated the use of LLMs to autonomously detect, diagnose, and recover from failures in workflow automation systems.
- Root cause analysis.
- Structured outputs.
- Safety constraints.
- Autonomous remediation.
- Reliability evaluation.
ExploringtheFrontiersofAI
My research bridges biologically inspired learning algorithms and production-ready AI systems. From alternatives to backpropagation to autonomous failure recovery, I explore ideas that push beyond the current state of the art.
Tools&TechnologiesIUse
Applied AI
Agentic Systems & Automation
Machine Learning
Backend
Infrastructure
Awards&Leadership
Awards
Hackatom Ethiopia 2024 Winner
2x AASTU Presidential Award
Dean's Award for Academic Excellence
Leadership
Technical Lead, GDG On Campus AASTU
Mentored 70+ student developers
ProfessionalDevelopment
Associate AI Engineer for Developers
DataCamp
Expected Completion: 2026
Interestedinthedetails?
Download my resume for a complete overview of my experience, projects, and technical skills.
Download ResumeLet'sBuildReliableAISystems
I'm open to opportunities involving Applied AI, Agentic Systems, RAG, and AI Automation.
Download ResumeLocation
Addis Ababa, Ethiopia