AI Engineer

AI Engineer Jobs

AI Engineers design, build, and deploy the models and systems that turn research into real products. You’ll own everything from data pipelines and training to inference APIs, evals, and monitoring. If you like shipping ML that actually gets used, this is the role.

Job Overview

AI Engineers bridge research and production. You’ll fine-tune LLMs, build RAG systems, optimize latency/cost at inference, and create eval frameworks that keep models reliable. Day to day = Python, PyTorch, vector DBs, and cloud infra. Success = models in prod serving millions with clear metrics for quality, safety, and ROI.

About the Company

We build AI-native products used by 50M+ people monthly for writing, coding, analysis, and automation. Our stack spans open models + proprietary research, with a strong emphasis on eval-driven development and user privacy. The team is senior, low-ego, and ships weekly. We invest heavily in GPUs, tooling, and research compute so you can move fast.

AI Engineer Responsibilities

• Build & ship models: Fine-tune, distill, or prompt-tune LLMs and multimodal models for specific product tasks
• Productionize ML: Package models into low-latency APIs with batching, caching, quantization, and fallbacks
• Data & eval: Design datasets, synthetic data pipelines, and offline/online evals for quality, safety, and grounding
• RAG & tools: Implement retrieval, reranking, chunking, and tool-use/agent architectures
• Infra & MLOps: Own CI/CD for models, feature stores, experiment tracking, and monitoring/drift detection
• Optimize: Profile and cut inference cost/latency with vLLM, TensorRT-LLM, ONNX, speculative decoding, etc.
• Cross-functional: Work with product, research, policy, and design to define success metrics and ship features
• Safety: Implement guardrails, red-team tests, refusal logic, and audit logging

Required Skills

• Strong coding: Python is default. Clean, tested code and PRs. Familiar with async, typing, and profiling
• ML fundamentals: Deep learning, transformers, loss functions, optimizers, regularization, and debugging overfits
• LLM practicals: Context windows, tokenization, KV-cache, function calling, embeddings, and eval methodology
• Data engineering: SQL, pandas, Spark or similar. Build repeatable datasets and labeling pipelines
• System design: Design for 100ms P99, autoscaling, multi-region, and cost constraints. Know tradeoffs
• Experimentation: Offline evals, A/B tests, metric design, and statistical rigor
• Communication: Explain model behavior to PMs, write RFCs, and document decisions

Qualifications & Experience

• Education: BS/MS in CS, EE, Stats, Math, or equivalent hands-on experience. PhD useful for research-heavy teams but not required.
• Experience: 3+ years building ML systems end-to-end, or 2+ years specifically on LLMs/generative AI in prod.
• Portfolio: Links to shipped models, open-source contributions, papers, or detailed write-ups of projects you owned.
• Nice to have: Experience with RLHF/DPO, multimodal models, agents, on-device ML, or CUDA kernels.

Technologies You’ll Work With

Salary & Benefits

Benefits: Medical/dental/vision 100% covered, 401k with 6% match, 4 months parental leave, unlimited PTO with 4 week minimum, $3k/yr learning budget, top-tier GPU access, and $10k home office stipend.

Why Join This Company?

Real users: Your work ships to millions, not internal dashboards.
Compute: Dedicated H100 clusters + open-model access. No waiting on resources.
Ownership: You own a surface area from data to monitoring, not just one layer.
Talent: Work with ex-DeepMind, OpenAI, and Anthropic engineers who publish and ship.
Mission: Build AI that’s useful, safe, and accessible. We measure success by helpfulness, not hype.

Application Process

1. Apply: Resume + GitHub or project that shows end-to-end ownership. Writing about tradeoffs helps.
2. Recruiter chat 30 min: Role scope, team fit, comp, and timeline.
3. Technical screen 60 min: Python + ML debugging. Expect a Colab on a real model issue.
4. Virtual onsite 4×45 min: ML system design, coding, eval design, and behavioral.
5. Offer: Target 5 business days post-onsite. We can do exploding offers if needed.

Related AI Jobs

• Research Engineer, LLMs: Focus on pretraining, new architectures, and long-context
• ML Platform Engineer: Build training/inference infra, not product models
• AI Product Manager: Define model UX, eval criteria, and ship roadmap
• Data Scientist, Generative AI: Metrics, experimentation, and user research on AI features
• AI Safety Engineer: Red-teaming, policy enforcement, and risk measurement

Similar Machine Learning Jobs

• MLOps Engineer: CI/CD, monitoring, and feature stores for all ML, not just genAI
• Applied Scientist: More modeling/research, fewer prod service responsibilities
• Data Engineer, ML: Focus on pipelines, labeling, and dataset quality at scale
• Computer Vision Engineer: Detection, segmentation, and multimodal model work
• Speech AI Engineer: ASR, TTS, and audio models for voice products

Frequently Asked Questions

• Do I need a PhD?
  No. We hire for proven ability to ship. A PhD helps for research-heavy roles, but most AI Engineers come from SWE or MLE backgrounds.
• PyTorch or TensorFlow?
  90% of our work is PyTorch. TF experience is fine, but you’ll use PyTorch day to day.
• How much is research vs engineering?
  ∼70% engineering: data, eval, serving, cost. ∼30% modeling: fine-tuning, prompting, architecture tweaks. If you want 100% research, look at Research Engineer.
• Remote?
  US/Canada remote is standard. We have hubs in SF, NYC, and Seattle with optional in-person weeks monthly.
• What’s the on-call like?
  One week every 8 weeks per team. We invest in reliability so pages are rare. You won’t be paged for training jobs.
• Can I work on open source?
  Yes, with approval. We contribute to vLLM, Transformers, and have our own OSS projects.