Uses
Languages
- Python — daily driver for everything ML
- TypeScript / JavaScript — web and tooling
- Java · C++ (basic)
ML / DL
- PyTorch — model training and research
- HuggingFace Transformers + Datasets — NLP / LLM work
- TensorFlow — when the job calls for it
- scikit-learn — classical ML and evaluation pipelines
- NumPy · Pandas · SciPy — data and numerical computing
- Weights & Biases · MLflow — experiment tracking
LLM / RAG / Agents
- LangGraph — stateful agent pipelines
- LangChain — RAG and retrieval tooling
- Anthropic / OpenAI / Groq APIs — inference and evaluation
- Qdrant — production vector search
- Neo4j — property graphs for GraphRAG
Infrastructure
- Docker + Docker Compose — containerisation
- FastAPI — backend APIs
- GNU/Linux — primary environment
- Slurm — HPC cluster job scheduling
- Git + GitLab CI / GitHub Actions — version control and CI
- pytest — testing
Editor
- VS Code — primary editor
- Claude Code — AI pair programming
Hardware
Go-to papers
- Attention Is All You Need — Vaswani et al.
- Matryoshka Representation Learning — Kusupati et al.
- ArcFace: Additive Angular Margin Loss — Deng et al.
- ModernBERT — Warner et al., 2024
