AI / Speech Modeling Software Engineer We are looking for a software engineer with strong interests in machine learning, data analysis, and mathematical modeling. This is a hands-on modeling role combining classical quantitative methods with modern AI, including statistical analysis, mathematical and physical modeling, signal processing, CNNs, RNNs, Transformers, representation learning, and foundation models. The goal is not simply to train neural networks or tune hyperparameters. We want to understand the underlying problem, extract structure from data, formulate useful models, test hypotheses, and choose the right level of model complexity. Depending on the problem, the right solution may be a neural network, a statistical or mathematical model, a signal-processing method, or a combination of approaches. What You Will Work On You will work primarily on speech and audio problems, with opportunities to work on broader quantitative modeling tasks.
Typical work includes:
Analyzing real-world datasets to identify patterns, structure, and useful signals Formulating problems as mathematical, statistical, or computational models Building models using physical intuition, signal structure, probability, optimization, and machine learning Developing speech and audio models using CNNs, RNNs, Transformers, and modern pretrained models Combining signal processing and classical modeling with modern AI approaches Designing experiments, evaluation methods, and model objectives Investigating model behavior and determining whether errors originate from data, assumptions, architecture, or optimization Building data pipelines for cleaning, transforming, visualizing, and analyzing data Translating successful modeling ideas into reliable software What We Are Looking For Strong coding skills, preferably Python and/or C++ Strong quantitative and analytical ability Solid foundations in linear algebra, probability, statistics, calculus, and optimization Experience with machine learning and neural networks Ability to analyze unfamiliar datasets and formulate models from real-world problems Ability to design experiments and reason independently from results Experience in speech, signal processing, time-series analysis, mathematical or physical modeling, or scientific computing is especially valuable. This position is open to candidates expecting either a Bachelor's degree or a PhD in Computer Science, Electrical Engineering, Applied Mathematics, Statistics, Physics, or a related quantitative field, in a US university. Candidates currently authorized to work in the U.S. under F-1 CPT or OPT are welcome to apply. Future employment-based visa sponsorship may be considered based on performance and business needs.
