This role pays more than
60%
of similar roles. Most pay
$175,750–$254,750
— the shaded band above.
At the midpoint, this role pays about
$225k
versus about
$215k
for comparable roles.
Based on 240 similar postings.
Employer
About Goldman Sachs
Goldman Sachs is a leading global investment banking, securities, and investment management firm providing financial services to corporations, financial institutions, governments, and individuals.
Goldman Sachs currently has
134 open roles
on FindRole.
Listed pay typically runs
$137,000–$250,000
across 55 roles with salary data.
VP AI Research/Applied AI - Deep Learning & Time Series serves as a hands-on individual contributor within the Engineering AI Research team, focusing on the intersection of frontier machine learning and quantitative finance. The role involves designing, training, and evaluating deep learning models for financial time series to solve problems involving non-stationary, low signal-to-noise data. Key responsibilities include building architectures like Transformers, CNNs, GANs, and diffusion models, while benchmarking them against classical econometric methods such as ARIMA and Kalman filters. The candidate will manage large-scale training on multi-node GPU clusters using PyTorch, TensorFlow, or JAX, and optimize inference via ONNX. Essential skills include Python proficiency, expertise in distributed training (DDP, FSDP), and a strong foundation in statistics and signal processing to develop reproducible models for market microstructure, cross-asset pricing, and macroeconomic indicators.
Design, implement, and train deep learning architectures like Transformers and GANs for financial time series forecasting.
Build and benchmark specialized sequence models against classical econometric methods such as ARIMA and Kalman filters.
Manage large-scale model training across multi-node GPU clusters using distributed data parallel (DDP) and FSDP techniques.
Develop robust evaluation frameworks including walk-forward cross-validation and uncertainty quantification for noisy financial data.
Partner with quantitative researchers to translate model outputs into risk-adjusted, capacity-aware signals for various asset classes.
Contribute reusable models, datasets, and tools to a firmwide shared research platform.
Mentor junior staff and present complex research findings to both technical and non-technical stakeholders.
Ensure all models comply with firm standards for risk management, data privacy, and ethical AI governance.
What we're looking for
Hold a Bachelor's, Master's, or Ph.D. in Computer Science, Machine Learning, Statistics, Mathematics, Physics, Electrical Engineering, Quantitative Finance, or a related field.
Possess at least 7 years of industry experience building and deploying deep learning models, specifically for sequential or time series data.
Demonstrate expertise in modern architectures including CNNs, Transformers, autoencoders, GANs, diffusion models, GNNs, Bayesian methods, and reinforcement learning.
Maintain a strong working knowledge of classical time series and econometric modeling such as ARIMA, GARCH, Kalman filters, and state space models.
Possess practical experience with specific time series architectures like WaveNet, N-BEATS/N-HiTS, DeepAR, PatchTST, or Time Series Foundation Models.
Exhibit expert-level Python skills and deep proficiency in PyTorch, TensorFlow/Keras, and/or JAX/Flax.
Demonstrate experience with distributed training (DDP, FSDP), multi-node GPU clusters, and model optimization workflows like ONNX.
Possess a rigorous foundation in statistics, probability, stochastic processes, optimization, and signal processing.