AI Engineer - Model Performance

Das ist der Job

ABOUT FATHOM We created Fathom to eliminate the needless overhead of meetings.

Darum lohnt es sich

Our AI assistant captures, summarizes, and organizes the key moments of your calls, so you and your team can stay fully present without sacrificing context or clarity. From instant, searchable call summaries to seamless CRM updates and team-wide sharing, Fathom transforms meetings from a source of friction into a place for alignment and momentum.

ROLE OVERVIEW We are hiring a Model Performance Engineer to own the speed, cost, and reliability of our model inference stack, and to build the fine-tuning infrastructure that makes the rest of the AI team faster. Experience building internal tooling that improves team velocity. A dynamic and collaborative engineering team.

Competitive compensation and benefits. ABOUT THE INTERVIEW You’ll meet the entire team. We’re a small company that creates magical experiences through the hard work of focused builders. We try to live our values - Care Deeply, Seek Leverage, Share Ownership, Sustain Urgency, and Be Tenacious - in everything we do, every day.

Role overview summary of the position is provided below. This is not a research role. You will optimize real systems serving millions of meetings, balancing throughput and latency considerations across hardware and software layers.

RESPONSIBILITIES Own inference performance: accelerate models and reduce cost through techniques such as quantization, batching strategies, GPU selection, and cold-start mitigation, while ensuring fast product experience during highly spiky traffic.

Own fine-tuning pipelines: build repeatable infrastructure so AI engineers can move quickly from dataset to deployed model, including distillation, adapters, and task-specific tuning.

Benchmark and evaluate deployment options: quantify tradeoffs across FP8 quantization, static vs dynamic quantization, serving frameworks, and compiler optimizations to meet performance targets. Debug and resolve production inference issues: trace regressions to serving framework updates or data handling paths and implement durable fixes.

REQUIREMENTS Hard Skills: Deep experience with LLM serving frameworks (e.g., vLLM, SGLang, TensorRT-LLM or similar) and tuning, including attention backends, scheduling, CUDA graph warmup, and prefix caching. Hands-on quantization experience (weight vs activation, per-channel vs per-tensor scaling, dynamic quantization considerations).

Production fine-tuning experience (LoRA/QLoRA SFT, familiarity with training frameworks, data formatting, learning rate schedules, and diagnosing training failures). Strong Python skills for writing serving infrastructure, benchmarking harnesses, and training pipelines.

Comfort with GPU profiling and performance analysis; ability to identify bottlenecks in compute, memory bandwidth, or scheduling overhead. Strong signal: Cost modeling for GPU infrastructure and ability to justify tradeoffs between GPU types. Experience with multimodal models and serverless GPU platforms.

Understanding of audio processing (codecs, chunking, sample rates). Not required: ML research background or publications. Prompt engineering expertise. Frontend or full-stack experience. Masters/PhD (though it’s fine if you have one). WHAT'S IN IT FOR YOU The opportunity to shape the foundational software services of a growing company.

A role that balances innovation and incremental improvement. A supportive environment that encourages innovation and personal growth. WHY YOU SHOULD JOIN US Opportunity for impact. We are ship-focused and still early enough for your work to have real effect. Startup experience.

Work closely with our CEO, a 2x founder/CEO with a background in computer science and product design. Remote-first. We schedule meetings sparingly and rely on async communication (Slack, Notion, Loom). Meet everyone you’ll be working with. Ask anything you like; we value transparency in the hiring process. Quick turnaround.

We typically move from start to finish in less than a week. HOW TO APPLY Include a brief write-up or demo of inference optimization or model serving work you’ve done. We care about the reasoning behind your decisions — why you chose a specific quantization strategy, how you diagnosed a performance regression, and what tradeoffs you navigated.

A GitHub repo, blog post, or a few paragraphs in your cover letter are suitable. #J-18808-Ljbffr