Software Engineer - Applied AI

Deepgram • San Francisco, CA

1d • $173,800 - $212,400 • Onsite

About The Position

Deepgram’s Applied AI team tackles voice AI problems without off-the-shelf solutions, working directly with enterprise customers and technology partners. The team’s work includes conversational AI and customer experience platforms, restaurant brands, and other projects that are often built from zero to one. Focus areas include robust speech recognition in noisy outdoor environments, self-service fine-tuning and model adaptation, domain-aware voice agents for drive-thru and phone ordering, agent scoring and evaluation, real-time operational analytics, and on-site hardware integration. The team works in person in San Francisco, ships to production quickly, and stays close to its customers.

The Software Engineer will ship production features for live voice agent products, own customer problems from diagnosis through production fixes, and help create reusable agent modules that enable faster launches. The role also contributes to improving accuracy on real-world menus and catalogs, fine-tuning and model adaptation, proactively identifying product problems, and running experiments to assess whether new functionality improves customer outcomes. Deepgram expects an AI-first approach to work: team members actively use and experiment with advanced AI tools, adapt quickly as technology changes, and apply AI effectively while maintaining quality.

Requirements

• Experience building and shipping backend services that run in production in the cloud.
• Uses modern AI coding tools as a default way of working, moving quickly without lowering the quality bar.
• Learns quickly and can take on unfamiliar parts of the stack with little ramp-up.
• Identifies problems, proposes solutions, and follows through without waiting for assigned work.
• Has taken something from an idea to production, professionally or independently.
• Communicates clearly with teammates and is comfortable working directly with customers.
• Prioritizes customer outcomes, quality, and avoiding over-engineering.
• Comfortable working across programming languages; the stack is mainly Kotlin and Python, with TypeScript on the front end.

Nice To Haves

• Hands-on experience training, fine-tuning, or adapting models through work, research, or substantial projects.
• Experience building evaluation or scoring systems for models or agents.
• Experience with audio, speech-to-text, or text-to-speech.
• Experience as a founder or early startup engineer, or taking a system from zero to one.
• Typically 2+ years of engineering experience, or equivalent demonstrated experience.

Responsibilities

• Ship production features for live voice agent products, such as drive-thru and phone ordering.
• Own customer problems end to end: understand the issue, identify its root cause, and deliver a fix to production.
• Build reusable agent modules, such as ordering and reservations, to help new customers and industries launch faster.
• Improve agent accuracy on real-world menus and catalogs and keep structured data consistent across locations.
• Contribute to fine-tuning and model adaptation for customer-specific vocabulary and domains.
• Proactively identify product problems, fix them, or propose solutions.
• Run experiments to determine whether new functionality improves customer outcomes.
• Use and experiment with advanced AI tools as a regular part of work, adapting to new models and capabilities.

What This Job Offers

• Job Type: Full-time
• Career Level: Mid Level
• Education Level: No Education Listed