Machine Learning Intern - KWS/AED

Summary Description:

Syntiant Corp., a leader in the high-growth AI software and semiconductor solutions space, is looking for a Machine Learning Intern to take on a critical role supporting our Algorithms team's work on keyword spotting (KWS) and audio event detection (AED) models deployed on ultra-low-power edge hardware.

The Machine Learning Intern will work alongside senior ML engineers to help build, evaluate, and improve deep learning models that run directly on Syntiant's NDP-class neural decision processors. These models must detect wake words, spoken commands, and acoustic events (e.g., glass breaking, alarms, sirens) in real time under extremely tight memory and power budgets. This role spans the full modeling pipeline, from signal processing and data curation through architecture design, training, and evaluation against hardware constraints.

Specific Duties and Responsibilities:

• Support development and evaluation of KWS and AED models, including single-stage and cascaded (multi-stage gate/verifier) detection architectures.
• Assist with audio pipeline and feature extraction work, including filterbank design, log-mel and PCEN-based frontends, and diagnosing numerical or performance issues in training/eval pipelines.
• Help design and prune CNN architectures to fit hardware constraints (fixed input shapes, 8-bit quantization, limited parameter budgets, restricted op sets such as depthwise separable convolutions with hardware-supported stride/pooling operations).
• Build and run false-accept (FA) diagnostic tooling, including categorized probe sets, confusion analysis, and Grad-CAM/occlusion-style visualization to understand what a model is actually keying on.
• Contribute to hard-negative mining and data augmentation strategies (e.g., SNR-based background noise mixing, targeted negative class collection) to reduce false accepts across everyday household/environmental sounds.
• Help plan and track data collection efforts, including structuring datasets by acoustic category/spec and maintaining collection logs and inventories.
• Analyze model run results across experiment variants (architecture, data, frontend) and summarize findings for the team.
• Collaborate with ML, DSP, and hardware/firmware engineers to validate models against real deployment conditions.

Qualifications, Education, and Experience Required:

• Candidate pursuing or has completed a Bachelor's or Master's degree in Computer Science, Electrical Engineering, Machine Learning, or a related field, with hands-on experience in deep learning for audio or speech (coursework, research, or project experience with CNNs/RNNs on spectrogram or time-series audio data).
• Proficiency in Python and a deep learning framework (TensorFlow/Keras preferred; PyTorch acceptable).
• Familiarity with audio signal processing fundamentals (spectrograms, mel filterbanks, feature extraction).
• Understanding of standard ML evaluation concepts (precision/recall trade-offs, ROC/DET curves, confusion analysis). Bonus if applied to detection/verification tasks rather than pure classification.
• Exposure to model efficiency concepts (quantization, parameter budgets, edge/embedded ML constraints) is a strong plus, though not required.
• Strong analytical mindset, comfort working with messy real-world data, and clear written communication for summarizing experimental results.
• Prior internship, research, or personal project experience in audio ML, KWS, or acoustic event detection is a plus but not required.

About Syntiant:

Founded in 2017 and headquartered in Irvine, Calif., Syntiant Corp. is a leader in delivering hardware and software solutions for edge AI deployment. The company's purpose-built silicon and hardware-agnostic models are being deployed globally to power edge AI speech, audio, sensor and vision applications across a wide range of consumer and industrial use cases, from earbuds to automobiles. Syntiant's advanced chip solutions merge deep learning with semiconductor design to produce ultra-low-power, high performance, deep neural network processors. Syntiant also provides compute-efficient software solutions with proprietary model architectures that enable world-leading inference speed and minimized memory footprint across a broad range of processors. The company is backed by several of the world's leading strategic and financial investors including Intel Capital, Microsoft's M12, Applied Ventures, Bosch Ventures, the Amazon Alexa Fund, and Atlantic Bridge Capital. More information on the company can be found by visiting www.syntiant.com.