Cognitiv is looking to enhance its Deep Learning Advertising Platform by improving key metrics in real-time bidding (RTB) algorithms and Large Language Model (LLM) integrations through machine learning research and optimization.
Requirements
- Strong grasp of architectures like Transformers and hands-on with PyTorch.
- Skilled in hyperparameter tuning, loss function design, and optimizing training pipelines.
- Proficient in Python for data manipulation and ML experimentation.
- Familiar with techniques like gradient boosting (XGBoost), PCA, and distributed training.
- Comfortable with AWS/GCP and have some exposure to Spark, Hadoop, or SQL.
- Worked with large language models or fine-tuned transformer-based architectures.
- Familiarity with real-time bidding (RTB) or online advertising models.
Responsibilities
- Optimize ML models – Improve predictive accuracy, inference speed, and efficiency in AdTech applications.
- Experiment & tune – Run hyperparameter tuning, explore new architectures, and fine-tune models to move the needle on key business metrics.
- Build strong datasets – Help construct and preprocess high-quality training data for ML pipelines.
- Run experiments – Conduct PyTorch experiments and evaluate results using clear, measurable metrics.
- Deploy at scale – Collaborate with researchers and engineers to build scalable ML pipelines for smooth deployment and iteration.
- Stay ahead of the curve – Keep up with deep learning research and propose novel approaches for model improvement.
- Communicate findings – Present research and experimental outcomes in clear reports that influence decision-making.
Other
- Pursuing or recently completed a Master’s or Ph.D. in CS, Stats, EE, or a related field.
- Can explain technical results clearly and thrive in a collaborative, fast-paced environment.
- Experience with C++ for ML model performance tuning.
- Hybrid work schedule of 3 days in office (Mon/Tue/Wed) and 2 days remote (Thursday/Friday).
- We are also open to remote applicants.
