Artificial Intelligence Intern - Remote

About the Company:

Abekus - a lean, fast-moving startup looking for AI interns who are curious, hands-on, and ready to apply AI in practical scenarios. You’ll work directly with the founding team and have a front-row seat to how real-world AI products are built.

Roles & Responsibilities:

• Contribute to the development and fine-tuning of ML models using Python and PyTorch.
• Experiment with modern LLMs and generative AI models to solve real product challenges.
• Work with a variety of tools and frameworks like Transformers, LangChain, OpenAI API, LlamaIndex, and others for tasks such as classification, generation, and retrieval.
• Help build and optimize training scripts, inference pipelines, and evaluation tools.
• Conduct literature reviews or repo deep-dives to bring new ideas from the rapidly evolving AI ecosystem into production.
• Assist in cleaning, preprocessing, and analyzing datasets.
• Work closely with engineers and founders to prototype and test AI-powered features.

Skills & Qualifications:

• Strong programming skills in Python.
• Good understanding of machine learning fundamentals (loss functions, metrics, overfitting, etc.).
• Some hands-on experience with deep learning libraries (e.g., PyTorch, TensorFlow).
• Familiarity with modern NLP and generative AI tools such as Hugging Face Transformers, OpenAI, Claude, LLaMA, LangChain, or similar.
• Awareness of ongoing trends in the AI/LLM space and eagerness to explore new tools and models.
• Self-motivated and proactive mindset — you’re not afraid to explore, experiment, and learn on the go.
• Bonus: You’ve worked on AI/ML side projects, GitHub repos, Kaggle competitions, or open-source contributions.

Benefits:

• Mentorship and direct collaboration with the founding team.
• Real-world experience applying AI to product use cases.
• Ownership of small but impactful features and experiments.
• A fully remote and async-friendly work culture.
• Opportunity for long-term collaboration based on performance.

Note: