Senior Research Engineer at AssemblyAI at AssemblyAI

Why AssemblyAI

AssemblyAI builds best-in-class Voice AI models, powering the next generation of voice applications used by thousands of customers including Granola, Fireflies, Figure AI, and CallRail. Our models process over 1 million hours of audio daily, handle more than 600 million inference calls monthly, and facilitate billions of end-user experiences. The Voice AI industry is currently at a critical inflection point, and we are actively seeking individuals passionate about joining a small team to help shape the future of this industry.

We are recognized as one of the most capital-efficient AI companies globally. With a team of fewer than 100 people, we generate approximately $600K ARR per employee, positioning us among the top 5 most revenue-dense teams within today's fastest-growing AI companies. This efficiency is a deliberate strategy to remain lean, move quickly, and empower every team member with significant ownership and impact. This presents a unique growth-stage opportunity where the business model is proven, the growth trajectory is steep, yet the team remains small enough for your direct contributions to be evident everywhere.

If you have experienced being bogged down by bureaucracy, lacked true ownership, or felt frustrated watching your work disappear into a slow-moving organization, AssemblyAI offers a refreshingly different environment. The company operates as a true meritocracy, characterized by minimal planning or approval processes and unrestricted access to the tools and information you need. For anyone genuinely invested in voice AI as a technology to build, rather than just a trend to follow, this is the place where the most interesting problems at the most interesting scale are being solved by a close-knit team.

We are committed to fostering an inclusive environment where our employees can bring their full selves to work and have equal opportunities to succeed. Regardless of your race, gender identity or expression, sexual orientation, religion, origin, ability, age, or veteran status, if this mission resonates with you, we encourage you to apply.

About the Role

We are looking for a Senior Research Engineer to join our Research team, where you will develop and enhance the systems that underpin large-scale distributed training, data processing, and inference. Our organizational goal is to rapidly solve customer problems and improve our products through efficient model development and measurement. The speed at which we achieve this depends on how quickly anyone can run an experiment, evaluate its results, and identify issues. Increasing this experimental velocity is central to this role.

You will be directly involved in and improving the very pipelines you work with. The ideal candidate will possess a deep understanding of modern deep learning systems, coupled with strong engineering expertise across JAX and TPUs, layer-level optimization, large-scale distributed training, streaming, low-latency and asynchronous inference, inference compilers, and advanced parallelization techniques. This is a cross-functional role requiring close collaboration with our researchers, infrastructure team, and production engineering. You will not simply be a handoff point but the person who acquires sufficient knowledge across each domain to follow problems through to their complete resolution.

At times, you will train models, conduct evaluations, and analyze data yourself to both deliver direct impact and determine what tooling or improvements would best amplify the team's efforts. We seek someone who understands the end-to-end impact they aim to create, measures it from the outset, and would prefer to discover they were wrong within a week rather than a quarter. This discipline is what transforms cross-functional ownership into a significant advantage. This role is embedded directly within the Research team.

What You’ll Do

• Elevate the team's experimental velocity, making it faster to launch experiments, obtain trustworthy results, and decide on next steps.
• Maintain and evolve our JAX training framework, ensuring its scalability and efficiency for large-scale distributed training runs on TPU infrastructure.
• Improve the quality of data that our models learn from by investigating quality issues, building tools to highlight them, and translating findings into measurable accuracy gains.
• Analyze the accuracy of production models, construct robust evaluation harnesses, and determine which improvements will yield the most significant impact for customers.
• Translate research prototypes into production-ready systems, involving the refactoring and modernization of model architectures and infrastructure throughout the process.
• Optimize production inference for speech language models, addressing both serving architecture and advanced techniques like quantization and speculative decoding.
• Investigate and resolve performance bottlenecks across the entire stack, from low-level kernels (XLA, Pallas) to high-level system design.
• Partner with researchers, infrastructure, and production engineering teams to trace problems to their ultimate source and implement durable fixes.

What You’ll Need

• Expert-level proficiency with JAX and TPUs, including the surrounding ecosystem such as Flax, Optax, and the XLA compilation pipeline.
• Demonstrated measurement discipline: you define what success looks like before starting, maintain skepticism of your own results until they are validated, and view an unexplained improvement as a problem rather than an automatic win.
• An appetite for the entire pipeline: while your core strength might be JAX and TPU performance, you are eager to investigate and resolve customer issues that trace back to data problems or evaluation blind spots. Successful individuals in this role typically start with deep expertise in one area and then continuously expand their knowledge outwards.
• Strong experience optimizing inference systems for production, ideally with large language models (LLMs) or speech models.
• Deep understanding of distributed training at scale, modern deep learning systems, and best practices in ML infrastructure.
• Familiarity with contemporary inference optimization techniques, including continuous batching, KV-cache management, sharding strategies, and quantization.
• Enthusiasm for refactoring and improving existing systems; you thrive on making products and code faster and better.
• Strong Python programming skills; experience with C++ or Rust for kernel-level work is a valuable asset.
• Excellent communication and a collaborative mindset, enabling you to clearly explain complex tradeoffs and prioritize high-impact work effectively.

Bonus Qualifications

• Domain knowledge in Speech-to-Text (ASR) technology, including ASR architectures, audio processing, and streaming inference.

Compensation Transparency

AssemblyAI is dedicated to recruiting and retaining exceptional talent from diverse backgrounds while ensuring pay equity for our team. Our salary ranges are established based on competitive rates for our size, stage, and industry, and represent one component of the comprehensive compensation, benefits, and other reward opportunities we provide.

Numerous factors contribute to salary determinations, including relevant experience, skill level, qualifications assessed during the interview process, and maintaining internal equity with peers on the team. The range provided below is a general expectation for this function as posted. However, we are open to considering candidates who may possess more or less experience than outlined in the job description, and any updates to the expected salary range will be communicated during the interview process. The specified range is the expected salary for candidates in the U.S. For candidates outside of the U.S., there may be a change in the range, which will be communicated throughout the interview process.

Salary range: $270,000 - $310,000 USD

Interview Process and Privacy Notices

• AI to Interview: If you are selected for an interview, please review this resource to better understand how AssemblyAI utilizes AI in our interview process.
• GDPR Privacy Notice: Candidates from the EU should review this job applicant privacy notice before applying.

Explore AssemblyAI

Learn more about AssemblyAI and our work:

• Speech-to-text
• Streaming speech-to-text
• Speech Understanding
• LLM Gateway
• Try the Playground
• Read about our $50M Series C fundraise
• Check us out on YouTube!