Engineering leader building practical AI for real products

San Francisco Bay Area • [email protected] • linkedin.com/in/sarahmsachs

<aside> ⛔

I am not currently open to new opportunities! Recruiters — please save my inbox 🙏

</aside>

<aside> 💚

Notion is hiring on the AI team. If you’re interested, email me at [email protected]

</aside>

About

Engineering leader focused on shipping practical, high‑leverage AI into real products at scale. We turn AI research into products that hundreds millions of people rely on every day. My work sits at the intersection of hands-on machine learning and organizational craft: standing up rigorous evaluation frameworks, embedding safety and reliability into the foundation, and partnering across product, design, infrastructure, legal, and go-to-market to carry AI from the lab to the real world. Our north star: make Notion the best system of record for AI work. The place every agent turns to for context, and the most capable, trustworthy agent for knowledge work itself.

Experience

• Eng Lead, AI — Notion

  Jul 2024 – Present • San Francisco Bay Area • On‑site

  Architecting the next generation of Notion AI: an agent that reasons over your work, finds what matters, and acts on it. I steer five pillars of our modeling organization:

  • Capabilities — expanding what the agent can do out of the box: new and open-source models, the agent harness, and intelligent use of compute.
  • Model Behavior — shaping how the agent reasons, responds, and earns trust, so its behavior is reliable, aligned, and feels right in real work.
  • Context — the right information at the right moment, fast and affordable: memory, high-performance search, and making Notion the system of record any agent can build on.
  • Agent Dev Velocity — the evaluation rigor and user-interaction analysis that let us move fast without breaking trust.
  • Platform — the durable foundation the whole stack compounds on: training, fine-tuning, and serving infrastructure for models at scale; the search and retrieval backbone; and the enterprise-grade security, privacy, and compliance that make all of it safe to trust with real work.

  Always hiring exceptional people across the org. If you want to help build AI that people genuinely trust, let's talk.

• Director of Engineering (AI and Infrastructure) — Tome

  Oct 2023 – Jun 2024

  • Led strategy and delivery for AI features used by 18M+ users in AI‑augmented presentations.
  • Built Autotome (end‑to‑end presentation generation), professional image generation, relevant web image search, and iterative chat‑based editing.
  • Owned OpenAI partnership, prompt engineering, evals, data pipelines, deployment, and cost and latency optimization, including fine‑tuned GPT‑3.5 on reserved capacity.
  • Developed open‑source model strategy; fine‑tuned and evaluated Llama‑7B and Mistral; trained on Mosaic, deployed on Fireworks; evaluated SDXL, DALL·E 3, and LoRAs for imaging.

• Head of NLP & GenAI; previously EM, ML & Data Science — Robinhood

  Sep 2020 – Oct 2023 • Menlo Park, CA

  • Set company GenAI strategy, partnered with OpenAI, and co‑led company‑wide AI hackathon.
  • Transitioned BERT‑powered CX chatbot into a guarded, compliant generative assistant for a regulated fintech context.
  • Managed 8+ ML engineers, data scientists, and analysts; led ML product, experimentation, and analytics for CX.
  • Built NLP content moderation for cash product; led routing, causal inference, and support automation.
  • Culture and people leadership: founded Women in Tech ERG, created ML Engineer leveling, interviews, and hiring rubric.

• ML Tech Lead, Founding MLE — Sunshine (formerly Lumi Labs)

  2018 – Sep 2020 • Palo Alto, CA

  • Focused on graph theory and link prediction; trained custom RoBERTa language models.

• Software Engineer, Machine Learning — Google

  2016 – 2018 • Mountain View, CA

  • Local Search and dining verticals; applied ML for personalization.
  • Launched and patented Personal Score in Google Maps, featured at Google I/O 2018 keynote.

Talks and Media

Some talks on what we’re building:

Talk 1

Talk 2

Some partnerships we are proud of: