Base Salary
$200k - $300k/yr
Responsibilities
- Build and improve production speech models and systems to enhance real-time conversational quality.
- Post-train LLMs for voice agents used in consumer finance.
- Develop evaluation systems, harnesses, and monitoring systems.
- Work across models, data, infrastructure, and product to improve end-to-end agent quality.
- Own major technical areas from problem definition through production deployment.
Requirements
- Strong engineering fundamentals and the ability to build reliable production systems.
- Experience with speech modeling or LLM post-training is sought, though prior experience is not required.
- Familiarity with ASR, TTS, turn detection, speech enhancement, multilingual speech, LLM post-training, or model evaluation.
- Ability to diagnose real-world model failures and turn them into practical improvements.
- High ownership, strong execution speed, and comfort working in ambiguous technical areas.
- Driven engineers with strong technical ability and enthusiasm for difficult AI problems are encouraged to apply.
Benefits
- In-person work in the San Francisco office 4 days per week with an 8:00 AM start time.
- Medical, dental, and vision coverage for full-time employees.
- Generous 401(k).
- Catered lunches.
Categories
About Salient
Salient enables US consumer lenders to increase recovery, resolve disputes and losses faster, reduce manual servicing workload, and strengthen exam readiness, all through AI agents purpose-built for regulated lending. Our platform uses compliance-first agents to automate complex workflows across collections, customer service, QA/compliance review, disputes and chargebacks, and total-loss mitigation. Every agent is designed around US supervisory expectations and uses borrower-level memory to deliver consistent, context-aware interactions. Built to integrate with the systems lenders already use, Salient helps banks, credit unions, auto lenders, and fintechs operate more efficiently without compromising compliance or borrower outcomes.