Applied Machine Learning Engineer

Nooks; United States; Remote; USD 80,000 - 120,000; Full time

Job summary

An innovative company is seeking a Machine Learning Engineer to tackle exciting challenges in AI-powered real-time collaboration. This role involves implementing machine learning features that enhance audio data processing, improve user interactions, and develop actionable strategies from conversation data. Join a team dedicated to pushing the boundaries of technology and making a significant impact in a rapidly evolving field. If you are passionate about machine learning and eager to contribute to groundbreaking projects, this opportunity is perfect for you.

Qualifications

• Experience in implementing ML features in production environments.
• Strong understanding of audio data and its processing techniques.

Responsibilities

• Implement ML features into the product focusing on audio data.
• Analyze conversation data to improve user experience and outcomes.

Skills

Machine Learning

Data Wrangling

Backend Engineering

GPT-3

Audio Signal Processing

Education

Bachelor's in Computer Science or related field

Master's in Machine Learning or related field

Tools

LLM embeddings

Probabilistic Models

Job description

The role

Note: Exact job title will be commensurate with experience

We have an ambitious product vision in a nascent area - AI-powered realtime collaboration - so there are a ton of interesting technical challenges on our roadmap. We’re hiring our first Machine Learning Engineer. This is a role focused on implementing ML features into Nooks. Our ideal candidate will have prior experience working in industry for a business where ML is a core part of the offering.

Examples of engineering problems you may work on

These are just examples, this list is non-exhaustive, and you don’t need experience in all of these areas. But hopefully you find some of them exciting!

1. Realtime audio AI & latency/precision/recall tradeoffs (algorithms & models)

   • We use audio data, transcription, silence detection, and several other signals to detect whether a live phone call is a voicemail, a human, or a dial tree. Latency is a critical factor alongside accuracy, as we need quick detection of humans. Our approach involves LLM embeddings, few-shot learning, data labeling, and continuous monitoring of model performance in production.

2. Smart call funnels & playbooks (data wrangling, backend engineering, GPT-3, UX)

   • Analyzing conversation data to identify points where representatives get stuck, and developing playbooks to guide them. Using GPT-3 and other LLMs to turn unstructured call data into actionable strategies and feedback loops.

3. Conversation embeddings & Markov models (ML modeling)

   • Understanding the structure of calls, predicting responses, and maximizing outcomes using conversation embeddings, clustering, and probabilistic models. Leveraging LLMs to generate embeddings, cluster similar conversations, and predict conversation trajectories.