Erik Arakelyan
Verified Expert in Engineering
Artificial Intelligence Engineer and Developer
Copenhagen, Denmark
Toptal member since April 8, 2022
Erik is an ML researcher currently pursuing a PhD in machine learning at the University of Copenhagen (UCPH). He specializes in topics such as NLP, Knowledge Graph optimizations, and explainability in NLP. Erik is looking for opportunities to apply his deep learning and software engineering skills in an exciting and challenging project.
Portfolio
Experience
- Software Engineering - 7 years
- Python 3 - 7 years
- Natural Language Processing (NLP) - 5 years
- Generative Pre-trained Transformers (GPT) - 5 years
- Optimization - 5 years
- TensorFlow - 5 years
- PyTorch - 4 years
- Knowledge Graphs - 4 years
Preferred Environment
Ubuntu, Atom, Visual Studio Code (VS Code), Python 3, PyTorch, TensorFlow
The most amazing...
...thing I've developed is an automated system for answering complex queries over Knowledge Graphs. It won the Best Paper award at ICLR 2021.
Work Experience
Visiting Researcher
Amazon Science
- Worked on grounded LLM reasoning and knowledge-based completion at Amazon Alexa.
- Created scalable pipelines for experimentation and inference over knowledge graphs.
- Researched multi-hop reasoning with retrieval-based augmentation over LLMs.
Machine Learning Engineer | Tech Lead
Arm
- Led a tech team in applied machine learning (AML) for tailoring deep learning (DL) models.
- Implemented model quantization and optimization pipelines and maintained the internally optimized model repo with complete CI/CD. This resulted in fourfold smaller and faster models across various architectures.
- Implemented ML pipelines as a part of the AML team. Researched efficient DL methods.
- Developed models for various NLP tasks and image processing.
Senior Data Scientist
Armenia National SDG Innovation Lab (UNDP)
- Implemented innovative solutions for improving public policy decision-making and created a platform for real-time analysis and prediction of current touristic activities in Armenia.
- Created DL models for continuous analysis of time series, images, and text.
- Created pipelines for continuous scraping and optimized flow for database management and ETL.
Teaching Associate
American University of Armenia
- Performed as a teaching associate of the deep learning and data structure course at AUA.
- Conducted weekly problem-solving sessions and programming labs.
- Directed a final project and bachelor capstone guidance. Composed and graded homework and exams.
Machine Learning Engineer
Teamable
- Created an end-to-end pipeline for automated CV parsing and analysis.
- Developed models for various NLP tasks like NER, semantic parsing, and topic detection.
- Developed models for image processing and Flask and Django apps for integration.
Software Engineering Consultant
Wolfram Research
- Created an end-to-end text to speech (TTS) pipeline integrated into Wolfram Mathematica.
- Developed software in the signal processing team, enhancing functions within Wolfram Mathematica.
- Created builds for different projects and optimized structure and flow in different projects.
Experience
Oral Outstanding Paper Award | ICLR 2021
https://arxiv.org/pdf/2011.03459.pdfWe proposed a framework for efficiently answering complex queries on incomplete Knowledge Graphs. We translated each question into an end-to-end differentiable objective, where a pre-trained neural link predictor computes the truth value of each atom.
Travelinsights
The tool allows for scalable real-time analysis of events, sentiments, topics, and insights into Armenia's touristic activities and trends.
Fifth Summer School on Mathematics and Applications at YSU
https://github.com/deeplanguageclassI created models for Armenian Transliteration and pipelines for Armenian Lemmatization and semantic segmentation.
Worked on improving word embeddings and held NLP workshops and sessions.
FastEnt
Created a pipeline for automated custom named entity recognition and disambiguation. Developed pipelines for continuous scraping and dataset generation and implemented robust methods for named entity generalization and detection.
TorchNorms
https://pypi.org/project/torchnorms/The library supports the seamless addition of new differentiable modules and supports complete CI/CD for safety.
Education
PhD in Artificial Intelligence
University of Copenhagen - Copenhagen, Denmark
Master's Degree in Artificial Intelligence
University College London - London, United Kingdom
Bachelor's Degree in Computer Science
American University of Armenia - Yerevan, Armenia
Skills
Libraries/APIs
TensorFlow, PyTorch, Keras, SpaCy, Pandas, llama.cpp
Tools
Jupyter, GitHub, DeepSeek, Mathematica, Atom, OpenAI o1
Languages
Python 3, SQL, Python, C++, R
Paradigms
Dynamic Programming, Linear Programming, Testing, ETL, Test-driven Development (TDD), Model Context Protocol (MCP), DevOps, Continuous Deployment, Continuous Integration (CI)
Platforms
Docker, Amazon Web Services (AWS), Ubuntu, Visual Studio Code (VS Code)
Storage
MongoDB, CouchDB, Data Pipelines, NoSQL
Frameworks
Flask, Chainer, Django, LlamaIndex
Other
Optimization, Software Engineering, Calculus, Linear Algebra, Machine Learning, Deep Learning, Natural Language Processing (NLP), Probabilistic Graphical Models, Knowledge Bases, Knowledge Graphs, Learning Transfer, Signal Processing, Cython, Reinforcement Learning, Open Source, Pipelines, Artificial Intelligence (AI), Data Science, Data Analytics, Data Analysis, Web Scraping, Audio, Document Parsing, NLU, Deep Neural Networks (DNNs), Large Language Models (LLMs), Language Models, Chatbots, Generative Pre-trained Transformers (GPT), Classification, Text Classification, Graphs, Retrieval-augmented Generation (RAG), Meta Llama, Mistral AI, OpenAI, Generative Artificial Intelligence (GenAI), Datasets, Vector Databases, Fine-tuning, Reinforcement Learning from Human Feedback (RLHF), LoRa, API Integration, AI Agents, Solution Architecture, Agentic AI, Agentic RAG Systems, Scalability, Semantic Search, FAISS, FastAPI, AI Pipeline, Research, Bayesian Inference & Modeling, Bayesian Statistics, Probability Theory, Statistics, Explainable Artificial Intelligence (XAI), Deep Reinforcement Learning, Image Processing, University Teaching, Servers, Networking, Amazon Mechanical Turk (MTurk), Algorithms, Statistical Methods, Causal Inference, Hardware Drivers, Fuzzy Logic, Torch, Deployment, Llama 3, Text Generation
How to Work with Toptal
Toptal matches you directly with global industry experts from our network in hours—not weeks or months.
Share your needs
Choose your talent
Start your risk-free talent trial
Top talent is in high demand.
Start hiring