Senior ML Engineer

Invoca•Los Angeles, CA

3d•$152,000 - $228,000•Remote

About The Position

AI is the product. The Context Engine and Invoca's agentic AI workflows only create customer value if the models behind them are fast, reliable, and continuously improving in production. This is a hands-on role focused on production leverage, not model experimentation in isolation. You will own the layer that turns trained models into dependable, high-throughput services that Data Scientists, Applied AI Engineers, and product teams build on every day. As a platform team, our real customers are the product teams who build on what we ship. Success in this role is measured by their outcomes: how quickly product teams can ship intelligent, differentiated customer experiences on top of the models and infrastructure you own, how reliably those experiences hold up as usage scales, and how much your work accelerates Invoca's ability to innovate with AI.

Requirements

5+ years of ML Engineering experience with a strong production focus
Advanced Python and deep learning proficiency (PyTorch, HuggingFace Transformers, spaCy)
Demonstrated track record deploying and maintaining transformer-based NLP models in production
Hands-on experience fine-tuning SLMs/LLMs (LoRA, QLoRA, PEFT) and optimizing models via quantization, batching, and throughput tuning
Proficiency with inference infrastructure: Triton, Baseten, vLLM, TGI, SageMaker, Vertex AI, or similar
Experience building production-grade APIs that expose ML models to downstream consumers
Familiarity with MLOps tooling, model monitoring, and eval platforms (Braintrust, MLflow, or equivalent)
B.S. in Computer Science, Engineering, Statistics, or equivalent; advanced degree a plus

Nice To Haves

Experience operating models in an agentic or multi-step AI workflow context is a strong plus
Familiarity with RLHF or preference training is a bonus

Responsibilities

Architect, implement, and maintain CI/CD pipelines for ML artifacts, including automated evaluation, versioning, and deployment.
Serve as the primary SME for operational excellence across the Invoca ML stack: uptime, latency, throughput, and cost.
Treat model serving as a product with real customers, the product teams who build on it, and hold yourself accountable to their ability to ship.
Own inference infrastructure end to end: model serving on Triton Inference Server, Baseten, and Kubernetes-based GPU infrastructure.
Profile and tune for low latency and high throughput; build robust, scalable APIs for internal and external model access.
Continuously look for step-function improvements in inference cost and speed, not just incremental tuning, so product teams never have to wait on the platform.
Apply parameter-efficient fine-tuning methods (LoRA, QLoRA, PEFT) to adapt transformer-based SLMs and LLMs for high-impact NLP applications in conversation intelligence.
Connect fine-tuning decisions directly to the customer experiences they enable, not just benchmark scores.
Contribute to model training infrastructure, data pipelines, and data lake foundations that keep the systems powering our models reliable and scalable.
Build evaluation and monitoring practices (Braintrust, MLflow, or equivalent) that catch regressions before customers do.
Partner closely with Data Scientists, Data Engineers, and Applied AI Engineers to build the foundational ML systems behind Invoca's agentic AI products.
Work directly with product teams to understand what they need to ship, and remove whatever is standing between them and shipping it.