Machine Learning Engineer
GuavaAbout the role
All roles
Engineering
Machine Learning Engineer
Full-Time
In-Office: Downtown Los Angeles, CA
Engineering
$140,000–$180,000 + equity
Apply for this role
Summary
Build the best tools in voice. You will work across the model lifecycle: designing experiments, training and evaluating models, and shipping them into a production system that handles real, regulated phone calls at scale. This is a hands-on engineering role. You will be close to the data, close to production, and close to the outcomes, because your models directly affect call quality, accuracy, and the trust our customers place in the platform.
What you'll do
Design, train, and evaluate ML models across Guava's platform: ASR, TTS, intent recognition, dialogue systems, summarization, and related NLP/NLU tasks.
Build and maintain LLM-based components in production: prompting, fine-tuning, structured outputs, tool calling, and retrieval where appropriate.
Pipeline and manage large, real-world datasets, including labeling workflows and data quality checks.
Take models from prototype to production: benchmark performance, validate against target accuracy, and ship with monitoring in place.
Partner with the Call Review and LLM working groups to root-cause production issues (ASR errors, hallucinations, latency, turn-taking) and turn them into model improvements.
Contribute to model governance: maintain the model inventory, document validation results, and support risk assessment for new deployments.
Work closely with platform and infrastructure engineers to optimize inference latency and cost at scale.
Stay current with ML/NLP research and bring practical, production-ready ideas back to the team.
What we're looking for
Strong computer science fundamentals: algorithms, data structures, and systems programming.
Solid math and statistics foundation: linear algebra, probability, and their application to machine learning.
Production experience building, training, or deploying ML models, ideally in NLP, speech, or signal processing.
Fluent in Python, with mature software engineering practice: clean code, testing, and a disciplined debugging loop.
Comfortable working with large datasets and modern numerical methods, including GPU-accelerated training.
Built something real with an LLM: structured outputs, tool calling, fine-tuning, or an LLM embedded in a production workflow.
Independent thinker who is comfortable owning a problem end to end, from data to deployed model.
Based in or willing to work from Guava's Downtown Los Angeles office.
Nice to haves
Experience with speech technologies specifically: ASR, TTS, speaker/voice biometrics, or telephony-adjacent audio processing.
Background in NLP: intent recognition, dialogue systems, summarization, or semantic search.
Experience with model governance, evaluation frameworks, or production ML monitoring.
Advanced degree (MS or PhD) in computer science, machine learning, or a related quantitative field.
Experience at an early-stage or fast-moving startup.
Why Guava
Work on models that run in live, regulated production, not a research sandbox.
Own real problems end to end, with direct access to senior engineering leadership.
Join a technical team with deep roots in speech, NLP, and applied ML.
Competitive base salary and early-stage equity.
Apply
Send a note and resume to hiring@goguava.ai. We read every application.
Apply for this role All roles
Benefits
Competitive base salary and early-stage equity.
Medical, dental, and vision coverage.
401(k) with employer match.
Flexible paid time off.
Paid parental leave.
Top-tier equipment and the tools you need to do your best work.
Guava is an equal opportunity employer. We celebrate diversity and are committed to creating an inclusive environment for all employees. We do not discriminate on the basis of race, color, religion, sex, sexual orientation, gender identity or expression, national origin, age, disability, veteran status, or any other protected characteristic. If you need a reasonable accommodation during the application or interview process, please let us know at hiring@goguava.ai.
In accordance with California pay-transparency law, the range shown on this posting reflects the pay we reasonably expect to offer for the role. Where a candidate lands within the range depends on job-related factors including experience, skills, and qualifications.
Applicants must be authorized to work in the United States. Research shows some candidates only apply when they meet every requirement. There is no perfect candidate. If you are excited about this role and believe you can bring value to the team, we encourage you to apply even if your experience does not line up with every point above.
Summary
Build the best tools in voice. You will work across the model lifecycle: designing experiments, training and evaluating models, and shipping them into a production system that handles real, regulated phone calls at scale. This is a hands-on engineering role. You will be close to the data, close to production, and close to the outcomes, because your models directly affect call quality, accuracy, and the trust our customers place in the platform.
What you'll do
Design, train, and evaluate ML models across Guava's platform: ASR, TTS, intent recognition, dialogue systems, summarization, and related NLP/NLU tasks.
Build and maintain LLM-based components in production: prompting, fine-tuning, structured outputs, tool calling, and retrieval where appropriate.
Pipeline and manage large, real-world datasets, including labeling workflows and data quality checks.
Take models from prototype to production: benchmark performance, validate against target accuracy, and ship with monitoring in place.
Partner with the Call Review and LLM working groups to root-cause production issues (ASR errors, hallucinations, latency, turn-taking) and turn them into model improvements.
Contribute to model governance: maintain the model inventory, document validation results, and support risk assessment for new deployments.
Work closely with platform and infrastructure engineers to optimize inference latency and cost at scale.
Stay current with ML/NLP research and bring practical, production-ready ideas back to the team.
What we're looking for
Strong computer science fundamentals: algorithms, data structures, and systems programming.
Solid math and statistics foundation: linear algebra, probability, and their application to machine learning.
Production experience building, training, or deploying ML models, ideally in NLP, speech, or signal processing.
Fluent in Python, with mature software engineering practice: clean code, testing, and a disciplined debugging loop.
Comfortable working with large datasets and modern numerical methods, including GPU-accelerated training.
Built something real with an LLM: structured outputs, tool calling, fine-tuning, or an LLM embedded in a production workflow.
Independent thinker who is comfortable owning a problem end to end, from data to deployed model.
Based in or willing to work from Guava's Downtown Los Angeles office.
Nice to haves
Experience with speech technologies specifically: ASR, TTS, speaker/voice biometrics, or telephony-adjacent audio processing.
Background in NLP: intent recognition, dialogue systems, summarization, or semantic search.
Experience with model governance, evaluation frameworks, or production ML monitoring.
Advanced degree (MS or PhD) in computer science, machine learning, or a related quantitative field.
Experience at an early-stage or fast-moving startup.
Why Guava
Work on models that run in live, regulated production, not a research sandbox.
Own real problems end to end, with direct access to senior engineering leadership.
Join a technical team with deep roots in speech, NLP, and applied ML.
Competitive base salary and early-stage equity.
Benefits
Competitive base salary and early-stage equity.
Medical, dental, and vision coverage.
401(k) with employer match.
Flexible paid time off.
Paid parental leave.
Top-tier equipment and the tools you need to do your best work.
Responsibilities
- Design, train, and evaluate ML models across Guava's platform: ASR, TTS, intent recognition, dialogue systems, summarization, and related NLP/NLU tasks.
- Build and maintain LLM-based components in production: prompting, fine-tuning, structured outputs, tool calling, and retrieval where appropriate.
- Pipeline and manage large, real-world datasets, including labeling workflows and data quality checks.
- Take models from prototype to production: benchmark performance, validate against target accuracy, and ship with monitoring in place.
- Partner with the Call Review and LLM working groups to root-cause production issues and turn them into model improvements.
- Contribute to model governance: maintain the model inventory, document validation results, and support risk assessment for new deployments.
- Work closely with platform and infrastructure engineers to optimize inference latency and cost at scale.
- Stay current with ML/NLP research and bring practical, production-ready ideas back to the team.
Qualifications
- Strong computer science fundamentals: algorithms, data structures, and systems programming.
- Solid math and statistics foundation: linear algebra, probability, and their application to machine learning.
- Production experience building, training, or deploying ML models, ideally in NLP, speech, or signal processing.
- Fluent in Python, with mature software engineering practice: clean code, testing, and a disciplined debugging loop.
- Comfortable working with large datasets and modern numerical methods, including GPU-accelerated training.
- Built something real with an LLM: structured outputs, tool calling, fine-tuning, or an LLM embedded in a production workflow.
- Independent thinker who is comfortable owning a problem end to end, from data to deployed model.
Benefits
- Competitive base salary and early-stage equity.
- Medical, dental, and vision coverage.
- 401(k) with employer match.
- Flexible paid time off.
- Paid parental leave.
- Top-tier equipment and the tools you need to do your best work.
Skills mentioned
About Guava
Your expertise just found its voice. Guava is conversational voice AI, shipped as an API. One proprietary stack, end to end: ASR, TTS and language models built together rather than stitched from third-party vendors, with real telephony from the first call. You already built the brain. Guava gives it a voice. Your logic runs the conversation in code, not prompts. Rules as code, versioned and tested and reviewed like the rest of your codebase. Your CRM, EHR and workflow engine stay the source of truth, and nothing is copied out. Prompt bloat is a maintenance nightmare. Change one line to fix one call and every other call becomes a regression test. Rules grow, confidence doesn't. Code scales. Security is not an add-on. Every call runs on the certified stack: SOC 2 Type II, HITRUST i1, PCI DSS Level 1, and a BAA available for healthcare deployments. Reports available under MNDA. 10B+ live agent minutes. 100+ deployments. 13 years in production. A 99.9% uptime SLA. 14 days from kickoff to your first live call. Where Guava ships: BPO, healthcare, vertical AI, insurance, fintech and government. Built for calls that have to be right.