Applied AI Engineer

Today we run thousands of conversations a day with real prospects. We need a harness that scores every change end to end, a benchmark suite that runs against any new model the day it drops, a red-team pipeline that probes our agents for failure modes, and self-improvement loops that feed production failures back into the eval set.

This is an evals and infrastructure role with deep LLM work. You will touch audio, but the center of gravity is the harness and the loops around it. Think of the harness as CI for voice conversations: it runs synthetic and real calls through our stack and scores agent behavior at every layer (STT, LLM, tools, TTS, full call outcomes), so we catch regressions before customers do. New models are coming out every few weeks, so the question is not just whether ours is good today, but whether we can tell within a week if a new open source release should replace it.

What you'll do

• Building and maintaining the eval framework that scores voice agent quality across transcription, LLM reasoning, tool use, TTS, and full-conversation outcomes
• Design voice agent behavior: system prompts, tool use, conversation flow, error recovery, and guardrails for real-time interactions
• Drive STT and TTS accuracy improvements by comparing providers, tuning configurations, and running rigorous A/B experiments the team can act on.
• Drive TTS quality improvements voice selection, latency vs. fidelity tradeoffs, prosody, edge cases
• Curate and grow our evaluation datasets, including hard-case mining from production traffic
• You'll build benchmarks we can run against any new model in days, run a red-team pipeline that probes for jailbreaks, hallucinated quotes, and compliance failures,
• Partner with backend engineers to wire eval signals into CI so regressions get caught before they ship
• Wire eval signals into CI so regressions block merges, and build self-improvement loops where hard cases from production auto-feed the eval set and our prompts optimize themselves over time.

What success looks like

Day 30

• You understand how our agents work across prompts, tools, evals, telephony, and customer systems.
• You have shipped a v1 of evals with at least one end-to-end metric the team trusts.
• You are sitting in on customer call reviews and tagging failure modes by hand to learn where the real problems live.
• You have one new model (open or closed) benchmarked against our production stack with numbers we can defend.

Day 60

• The eval system runs on updates and blocks merges that regress on a known set of cases.
• We have a first red-team suite covering at least three classes of failure modes (jailbreaks, hallucinated quotes, compliance), running on a schedule.
• Hard-case mining from production calls is automated, so the eval set grows without anyone triaging every example by hand.
• At least one open source model (Qwen, DeepSeek, or similar) is benchmarked against our production stack with a defensible recommendation on whether to switch.

Day 90

• We can swap in any new LLM and have a numbers-backed answer on whether to ship it within a week.
• DSPy or GEPA-style prompt optimization is running over at least one production voice flow, and you have shown measurable lift.
• Self-improvement v1 is live for at least one failure pattern. The same problem does not get solved twice because the system feeds the fix back into the platform.
• You are spotting failure patterns across customer accounts and turning them into product fixes the rest of the team builds on.

Must-haves

• ML engineering experience shipping production systems
• Strong Python and a working ML stack (PyTorch, Huggingface, pandas, scikit-learn)
• Hands-on experience designing LLM-based agents: prompting, tool/function calling, multi-turn state, structured outputs
• Hands-on experience building evals or eval frameworks for ML, LLM, or voice systems. Built LLM-as-judge eval pipelines and know their failure modes
• Practical experience with ASR/STT comparing providers, fine-tuning, or running open models like Whisper
• Practical experience with TTS systems (ElevenLabs or open models)
• Comfortable working with audio data: sample rates, codecs, noise, alignment

Nice-to-haves

• Designed voice agents specifically handled barge-in, interruption recovery, disfluencies, and natural turn-taking at the prompt/behavior layer
• Experience with diarization, VAD, or endpointing models
• Audio dataset curation, labeling, or annotation pipelines
• Trained or fine-tuned ASR or TTS models from scratch or on domain audio
• Experience with active learning or data-flywheel patterns over production traffic
• Open-source contributions to AI/ML frameworks
• Familiarity with cost/latency tradeoffs across model providers for real-time voice

A meeting lands on your calendar

If they’re a yes, I book the chat. You show up — that’s the whole job-hunt.