AppFolio Inc is building voice and conversational AI agents inside its Realm-X AI-native platform, with a focus on reliable real-world experiences. This Senior Machine Learning Engineer role is based in Atlanta, GA (onsite) and centers on production-ready agent pipelines, low-latency streaming infrastructure, and rigorous agent quality evaluation.
You will work on multi-turn, real-time voice and chat interactions, making system-level decisions that balance reasoning quality with latency and cost. You’ll also lead a small pod to strengthen observability, incident response, and evaluation practices as the product scales across AppFolio’s customer base.
What you’ll do
- Architect and ship voice and text agent pipelines for real-time, multi-turn customer interactions.
- Make principled trade-offs between reasoning depth and latency across frontier LLMs, smaller models, and routing strategies.
- Lead a small pod of ML and platform engineers, raising the bar on agent evaluation, observability, and incident response.
- Partner with Product and Voice channel teams to define KPIs, eval harnesses, and acceptance criteria for agent quality.
- Drive selective Small Language Model (SLM) fine-tuning and inference optimization to improve voice latency and cost.
- Translate agent and voice research into production systems that deliver low-latency, multi-channel experiences at scale.
What you bring
- 5+ years of experience developing and scaling web-based applications, preferably in a SaaS environment.
- Experience shipping production AI agents that serve real users in voice and/or text channels.
- Strong systems thinking, focused on pipelines and system design rather than only model work.
- Sound engineering judgment and ability to maintain high standards.
- Humble, collaborative, low-ego approach and a track record of elevating others.
- Value work-life balance as a foundation for sustained high performance.
- Deep, shipped experience with LangChain, LangGraph, LangSmith, and LangChain Deep Agents (or equivalent agent frameworks).
- Hands-on experience with Voice-to-Voice models plus traditional TTS/STT pipelines, including trade-off understanding between end-to-end voice models and modular STT + LLM + TTS architectures.
- Strong grasp of LLM reasoning behavior, tool use, structured output, and reasoning-versus-latency trade-offs across providers.
- Production experience with Twilio (or comparable telephony) and AWS.
- Expert Python, async programming, and WebSockets for real-time bidirectional streaming.
- Solid foundation in deep learning, model evaluation, and inference optimization, including deployment with Docker on AWS.
- Proven ability to lead a small team, mentor engineers, and partner credibly with Product and Design.
Nice to have
- Bachelor’s, Master’s, or Ph.D. in Computer Science or a related technical discipline.
- Experience fine-tuning SLMs for domain-specific voice applications.
- Familiarity with RAG over structured business data and tool-using agents over API surfaces.
- Experience in regulated or customer-facing industries with strict reliability requirements.
- Publicly verifiable work on GitHub, in open-source agent frameworks, or community competitions.
Compensation & benefits
- Base pay: $167,200 - $229,900 per year.
- Actual compensation will be determined based on factors including skills, education, experience, and internal equity.
- Benefits: Regular full-time employees are eligible; see here.
Location & interview process
This role is onsite in Atlanta, GA. Find out more about locations by visiting the company site. All late-stage candidates complete an in-person meeting with an AppFolian as part of the hiring process.
Technologies: Python, async programming, WebSockets, LangChain, LangGraph, LangSmith, LangChain Deep Agents, Voice-to-Voice models, TTS, STT, LLM, Small Language Models (SLM), Twilio, AWS, Docker, real-time, bidirectional streaming.
