Machine Learning Engineer

About Deepdesk

Deepdesk builds real-time AI assistance that empowers customer service agents across chat, email, and voice. Our Agent Assist solution leverages cutting-edge machine learning, NLP, and voice technologies to deliver instant suggestions, rewriting, context understanding, and emerging agentic AI capabilities.

We are shaping the next generation of intelligent customer support through fast, production-grade systems that work seamlessly across multilingual and omnichannel environments.

Role Overview

We’re looking for a Machine Learning Engineer to develop and optimise the ML intelligence behind Deepdesk’s Agent Assist platform. This role is perfect for someone who thrives in fast-paced, production environments and wants to work on impactful, real-time AI systems.

You’ll design and deploy ML/NLP models, enhance voice/STT pipelines, and collaborate closely with engineering to push the boundaries of agentic AI. Your work will directly improve how customer service agents communicate across chat, email, and voice.

Key Responsibilities

• Build and optimise ML and NLP models for real-time agent assist.
• Develop algorithms for search, autocomplete, ranking, and rewriting.
• Implement and refine multilingual and omnichannel (text + voice) capabilities.
• Integrate, tune, and deploy Speech-to-Text (STT) pipelines for voice-based use cases.
• Run experiments, model evaluations, and performance tuning.
• Design scalable ML infrastructure, monitoring, and production-ready components.
• Work closely with engineering to shape new agentic AI workflows.

Must-haves

• Strong Python development and algorithmic skills.
• Experience in ML or NLP (e.g., embeddings, classification, transformers).
• Hands-on expertise with PyTorch or TensorFlow.
• Ability to build clean, reliable, production-grade ML components.

Nice-to-haves

• Experience with voice/STT models (Whisper, wav2vec2, DeepSpeech, etc.).
• Background in agent assist, conversational AI, or agentic AI systems.
• Familiarity with Kubeflow, MLOps, or Google Cloud Platform (GCP).

Tech Stack

Python · TensorFlow · PyTorch · Scikit-learn · SpaCy · Kubeflow · GCP · Whisper / wav2vec2