Sr AI/ML Speech R&D Scientist

The future is what you make it.We are seeking an experienced Senior Artificial Intelligence & Machine Learning Specialist with expertise in speech processing to contribute to real-time speech anonymization solutions. You will be responsible for developing, training, and optimizing machine learning models aimed at transforming speech signals in real-time, ensuring the de-identification of speakers. This role will require you to work with large datasets, implement cutting-edge AI/ML techniques, and build systems optimized for high performance in CPU/GPU environments. You will be on a R&D team that works with engineers, researchers, and product managers across Honeywell to design and evaluate state-of-the-art product concepts. While this opening emphasizes speech technology skills, future R&D efforts may require translating skills to other technology areas (e.g., biosensing, augmented reality / virtual reality / near-to-eye, aircraft flight deck automation).LOCATION: The preferred location for this role is Deer Valley, Phoenix, AZ. The Plymouth, MN location will also be considered if needed. Are you ready to make the future with us?When you join Honeywell, you become a member of our global team of thinkers, innovators, dreamers and doers who make the things that make the future.BENEFITS: Benefits provided may differ by role and location. Learn more at benefits.honeywell.com.Unlimited Vacation Plan with No Preset Maximums Flexible Hybrid Work Schedule Medical/Rx Health Savings Account (HSA)Dental/Vision Short/Long-Term DisabilityEmployee Assistance Program (EAP)401(k) Plan Education AssistanceKEY RESPONSIBILITIES:Develop advanced speech transformation and speech pseudo-generation models.Design and implement machine learning architectures such as complex multi-head attention transformers, variational autoencoders (VAEs), and RNN/LSTM/CNN-based models for speech anonymization.Optimize AI models for real-time performance using model quantization, and model compression.Handle streaming audio processing with Voice Activity Detection (VAD) and implement models that operate in real-time.Develop and optimize AI models for both CPU and GPU environments (CUDA, cuDNN) using frameworks like TensorFlow and PyTorch.Train AI models on large-scale datasets and improve performance using automated platforms for testing and training.Collaborate with cross-functional teams to integrate AI solutions with the software architecture.Conceptualize and conduct applied research figuring out research developments best suited to solve open-ended problems.Respond to government funding solicitations and perform on government research contracts.Submit patent applications, academic publications, and share technical knowledge.U.S. PERSON REQUIREMENTS:Due to compliance with U.S. export control laws and regulations, candidate must be a U.S. Person, which is defined as, a U.S. citizen, a U.S. permanent resident, or have protected status in the U.S. under asylum or refugee status or have the ability to obtain an export authorizationYOU MUST HAVE:Bachelor’s degree from an accredited institution in a technical discipline such as the sciences, technology, engineering or mathematics6+ years of experience in developing AI/ML models for speech/audio processing with a deep expertise in CPU/GPU parallelism, multi-threading, real-time computing, and streaming data.WE VALUE:Master’s or Ph.D. in Computer Science, AI, Machine Learning, or a related field.Deep expertise in AI/ML with experience in speech processing tasks (e.g., speech synthesis, speech de-identification, speaker recognition, voice conversion, accent conversion, noise handling).Proficient in Python, with hands-on experience in PyTorch and TensorFlow.Experience with model optimization, including quantization, pruning, and compression.Strong knowledge of CUDA, cuDNN, and GPU-based deep learning optimizations.Speech processing experience with hybrid or End-to-end techniques.Handling complete code from data pre-processing to model deployment.Proven track record of developing speech solutions in a real-time and noisy production environment.Ability to handle noisy data in speech processing and improve the robustness of AI models.Familiarity with variational autoencoders (VAE) and transformer-based architectures.Experience with Linux and iOS environments.Familiarity with speech naturalness detection, emotion extraction, and accent identification.Experience working on speech anonymization and pseudo speech generation.Knowledge of voice accent conversion techniques and speaker ID systems.Exposure to speech enhancement and noise cancellation techniques.Strong history of academic publications on speech technology.Strong skills in proposal writing.Honeywell is an equal opportunity employer. Qualified applicants will be considered without regard to age, race, creed, color, national origin, ancestry, marital status, affectional or sexual orientation, gender identity or expression, disability, nationality, sex, religion, or veteran status.