Machine Learning Engineer Intern – Voice AI

ALEBEX AI is a Canadian-owned AI company based in Vancouver, BC, focused on developing an agentic workforce platform where AI agents perform real operational work — including answering calls, qualifying leads, and executing multi-step tasks — under human supervision.

The platform enables AI agents to operate across voice and digital channels, combining real-time communication, proprietary voice technology, AI models, integrations, and orchestration to execute real business workflows.

It operates on Canadian infrastructure, including on-premises AI inference, and is currently used in production by real clients.

ALEBEX is a small, fast-moving team focused on developing practical AI solutions for real-world business applications.

Its technology is built for production use, with active customers and operational workflows rather than demonstrations or experimental prototypes.

The successful candidate will work with the ALEBEX AI team on the development, evaluation, and optimization of machine learning models powering its Voice AI platform as part of this internship.

CAD 30 per hour

: Monday – Friday; Full-time preferred.

• Currently enrolled in a co-op-eligible program in Computer Science, Machine Learning, Artificial Intelligence, Electrical Engineering, or a related STEM field.
• Strong Python programming skills.
• Familiarity with PyTorch and modern deep learning workflows.
• Hands-on experience with at least some of the following:
• Supervised Fine-Tuning (SFT).
• LoRA, QLoRA, PEFT, or related post-training techniques.
• Hugging Face Transformers.
• LLM or speech model fine-tuning.
• GPU-based model inference.
• Online or real-time model serving.
• Good understanding of transformer architectures and modern LLMs.
• Comfortable working in a Linux environment.
• Strong problem-solving skills and willingness to learn.

• Experience with Voice AI, ASR, TTS, VAD, turn detection, or speech-language models.
• Experience deploying models with vLLM, TensorRT-LLM, Triton, ONNX Runtime, or similar inference frameworks.
• Understanding of KV caching, batching, quantization, prefix caching, or speculative decoding.
• Experience with Docker and cloud GPU environments such as AWS.
• Experience evaluating latency-sensitive machine learning systems.

• Fine-tune LLMs and speech models using SFT, LoRA, QLoRA, PEFT, and related post-training techniques.
• Experiment with training strategies, model configurations, and datasets.
• Support the development and improvement of models for Voice AI applications.

• Build and improve models for ASR, turn detection, interruption handling, intent understanding, and other Voice AI tasks.
• Work with speech, language, and speech-language models used within real-time voice agents.
• Help evaluate model behaviour across different voice and conversational scenarios.

• Prepare, clean, organize, and validate datasets used for model training and evaluation.
• Build and maintain evaluation pipelines for machine learning models.
• Benchmark models across accuracy, latency, resource usage, and other relevant metrics.

• Deploy machine learning models for online and real-time inference.
• Optimize inference latency, throughput, GPU memory usage, and concurrency.
• Experiment with batching, caching, quantization, and other inference optimization techniques.

• Work with inference frameworks such as vLLM, Hugging Face Transformers, PyTorch, ONNX Runtime, or similar technologies.
• Help integrate machine learning models into ALEBEX's real-time Voice AI infrastructure.
• Support model serving, deployment, monitoring, and performance testing.

• Experiment with model architectures, training approaches, and inference optimizations.
• Investigate technical problems and contribute practical solutions.
• Document experiments, findings, benchmarks, and implementation decisions.

• Help develop machine learning systems that operate under real-world production constraints
• Evaluate trade-offs between model accuracy, latency, reliability, GPU resource usage, and cost.
• Monitor and troubleshoot model behaviour in production inference environments.

This co-op placement involves access to sensitive and confidential information.

The successful candidate will be required to sign a Non-Disclosure Agreement (NDA) and Confidentiality Agreement prior to commencing the term.

Covered materials may include:

• Client information, configurations, and business data
• AI agent conversations, workflows, and performance data
• Proprietary software, source code, infrastructure, and architecture
• Internal product strategy and development materials
• Unreleased product features and technology
• Customer integrations and technical documentation

All confidential information must be handled in accordance with applicable Canadian privacy legislation, Alexander Innovation Centre policies, and ALEBEX's internal data governance and security requirements.

Email your application to careers@alebex.ai, using the subject line:

"[Your Full Name]_Machine Learning Engineer Intern (Voice AI)"

Please include:

• Resume/CV or LinkedIn profile.
• A link to something you have built, such as GitHub, a live project, or a short screen recording. We would rather see one meaningful project than a long list of projects.

Only shortlisted applicants will be contacted.

No phone calls please.

*All qualified candidates are encouraged to apply; however, Canadians and permanent residents will be given priority.

Thank you!*