Duties and Responsibilities
Specific Tasks
- Work with large language models (LLMs) to generate insights from user data (e.g. summarising behaviour, categorising goals).
- Fine-tune and optimise pre-trained models (e.g. decision trees, CNNs) for tasks like attention estimation and distraction detection.
- Build pipelines to clean, label, and process behavioural data from Focus Bear’s mobile, desktop, and web apps.
- Write production-ready Python code to deploy and monitor models in live environments.
- Collaborate with developers and product managers to turn models into features users actually interact with.
Key Objectives
- Experiment with prompt engineering and fine-tuning for LLM-based features.
- Optimise model performance for speed, memory, and accuracy — especially for mobile or edge contexts.
- Evaluate model outputs for alignment with real-world user experiences.
- Improve and maintain internal tools for model monitoring, testing, and retraining.
Position Benefits
- Get hands-on with the latest in LLMs, lightweight ML, and behavioural modelling.
- Build real features that improve executive functioning for neurodivergent users.
- Learn how to take ML models from Jupyter notebook to production API — with real users depending on your work.
- Collaborate with a purpose-driven team tackling real-world mental health challenges.
Team Description
- You’ll join a collaborative product + engineering team building tools that actually help people (not just optimise click-through rates).
- We care about ethical AI, fast feedback loops, and making our tech accessible to people with ADHD and Autism.
- Expect daily check-ins, sprint-based work, and lots of feedback and support.
- We use GitHub, Discord, Python (Pydantic, FastAPI), Postgres, and Render/AWS.
Skills Required
- Python + familiarity with ML libraries (e.g. scikit-learn, PyTorch, or HuggingFace Transformers).
- Comfort working with structured and unstructured data.
- Interest in LLMs, prompt design, and efficient model deployment.
- (Bonus) Some understanding of CNNs, tree-based models, or time-series analysis.
- (Bonus) Experience with APIs or containerisation (Docker, FastAPI).
Typical Week
You’ll spend your time designing prompts, fine-tuning models, evaluating outputs, and collaborating with devs to ship updates. Some days you’ll prototype new features; others you’ll write tests, run experiments, or polish pipelines so they don’t break in production.
Skills Development
You'll gain hands-on experience with the full ML pipeline — from prompt tuning and model selection to integration and monitoring. You'll learn how to work with production LLMs and refine classic ML models to handle real, noisy data from thousands of users.
Internship Nature
This is a hands-on, applied ML internship ideal for students in AI, data science, or software engineering. You’ll work on real-world challenges, contribute to meaningful tech, and learn what it takes to ship reliable ML-powered features in a production environment.
