Machine Learning Research Engineer

hackajob on-Demand is currently partnering with an AI startup company to help them hire the best talent. At on-demand, we match and speak with exceptional talent like you and provide insights into the problem they are looking to solve and the interview process.

Role: Machine Learning Research Engineer

Opportunity: Perm or Contract

Based: London or New York (remote possible but ideally onsite in either city)

About Us

We are a stealth-mode startup developing cutting-edge AI and machine learning tools for the financial sector. Our mission is to revolutionize how hedge funds leverage advanced technologies for data analysis and decision-making. We're building a diverse team of experts from various fields to create innovative solutions that push the boundaries of what's possible in financial technology.

The Role

We're seeking an exceptional Machine Learning Research Engineer to join our core team. You'll work on developing novel AI models and architectures, with a particular focus on adapting and repurposing large language models (LLMs) for complex financial data analysis and prediction tasks.

Key Responsibilities

• Collaborate with our research team to conceptualize and implement novel machine learning approaches for financial data analysis

• Develop and optimize complex neural network architectures, including modifications to state-of-the-art LLMs

• Implement efficient tensor operations and custom layers for GPU acceleration

• Design and conduct experiments to validate new model architectures and approaches

• Contribute to our proprietary AI framework and tools

• Stay current with the latest advancements in machine learning research and identify potential applications for our products

Requirements

• PhD or equivalent experience in Machine Learning, Computer Science, or a related field

• Strong understanding of deep learning architectures, especially transformers, LSTMs, and their variants

• Expertise in implementing and optimizing complex neural networks using PyTorch, TensorFlow, or JAX

• Proficiency in Python and C++, with experience in CUDA programming for GPU optimization

• Solid understanding of linear algebra, calculus, and other mathematical foundations of machine learning

• Experience with distributed computing and model parallelism for large-scale model training

• Strong software engineering skills, including version control, testing, and CI/CD practices

• Excellent problem-solving skills and attention to detail

• Ability to communicate complex technical concepts clearly and work collaboratively in a research-driven environment

• Deep understanding of classical machine learning algorithms (e.g., SVM, Random Forests, Gradient Boosting)

• Comprehensive knowledge of NLP techniques, including tokenization, embeddings, and language modeling

• Strong grasp of the inner workings of LLMs, including:

• Transformer architecture and its variants (e.g., GPT, BERT, T5)

• Attention mechanisms and their optimizations

• Pre-training and fine-tuning techniques

• Prompt engineering and few-shot learning

• Model compression techniques (e.g., quantization, pruning, distillation)

• Familiarity with LLM training and inference optimizations (e.g., mixed-precision training, efficient attention)

• Understanding of current limitations and challenges in LLMs (e.g., hallucinations, bias, long-range dependencies)

Preferred Qualifications

• Published research in top-tier ML conferences (NeurIPS, ICML, ICLR) or journals, especially in areas related to LLMs or financial applications of ML

• Experience applying ML to complex real-world problems in industries such as healthcare, autonomous systems, energy, finance, or scientific computing

• Familiarity with reinforcement learning and its applications to dynamic systems

• Contributions to open-source machine learning projects, particularly those involving LLMs or NLP

• Experience with high-performance computing in data-intensive fields

• Knowledge of financial markets and quantitative finance concepts (a plus, but not required)