Role Overview
- Design and train supervised and unsupervised models
- Develop NLP and LLM solutions
- Perform fine-tuning of foundational models
- Implement RAG (Retrieval-Augmented Generation) architectures
- Create training, validation, and inference pipelines
- Evaluate model performance (accuracy, F1, ROC, BLEU, etc.)
- Implement monitoring for drift and performance degradation
- Collaborate with engineering teams to deploy models to production
- Manage computational costs for training and inference
Requirements
- Languages
- Python (primary)
- SQL (analysis and feature engineering)
Machine Learning & Deep Learning
- PyTorch
- TensorFlow
- Scikit-learn
- XGBoost / LightGBM
LLM & GenAI
- OpenAI APIs
- Open-source models via Hugging Face
- Transformers
- Fine-tuning (LoRA, PEFT)
- Advanced prompt engineering
- Embeddings and vector similarity
- RAG architectures
MLOps
- MLflow
- Weights & Biases
- Docker
- CI/CD for models
- Model monitoring
- Dataset versioning
Infrastructure
- GPU environments
- Kubernetes
- Cloud (preferably one of the following):
- Amazon Web Services
- Google Cloud
- Microsoft Azure
Hard Skills
- Applied statistics and probability
- Predictive modeling
- NLP
- Model evaluation and validation
- Advanced feature engineering
- Fine-tuning LLMs
- Building ML pipelines
- MLOps and model deployment
- Performance and cost optimization
Soft Skills
- Deep analytical thinking
- Experimental mindset (constant hypothesis testing)
- Clear technical communication
- Ability to translate business problems into mathematical models
- Technical autonomy
- Sense of responsibility for model impact
Tech Stack
- AWS
- Azure
- Cloud
- Docker
- Kubernetes
- Python
- PyTorch
- Scikit-Learn
- SQL
- Tensorflow
Benefits
- Position also open to candidates with disabilities (PwD)