Head of AI

Who we are at Osmo:

Osmo is a digital olfaction company, on a mission to give computers a sense of smell to improve the health and wellbeing of human life. Why? Our sense of smell both enriches and saves lives, and has a deep and direct connection to our emotions and memory.

Olfactory Intelligence has applications across industries including fragrance, manufacturing, security, medicine, and more. We believe in the power of automation and thoughtfully applied AI/ML to solve problems beyond the reach of human intuition alone.

Osmo is building the first computer that can smell. Olfactory Intelligence (OI) is the model suite behind that: it predicts how a molecule or a mixture will smell, generates new formulas from a creative brief, and now powers commercial products for some of the largest consumer goods companies in the world. It runs on 5.4M+ human sniff annotations, 100M+ mass spectral readings, and a molecule library in the billions, data no one else has.

We are hiring a Head of Applied AI to own the team that builds these models. You will lead the ML engineers and research scientists, set the roadmap and the quality bar, and own the path from research notebook to production API. You will report to the CTO and work daily with Software Platform, Product Engineering, Chemistry, Sensory Science, and the perfumers and evaluators who are the ground truth for every prediction. This is a hands-on leadership seat at a company of roughly 100 people. You will read the code, run experiments, and be held to hard contractual accuracy gates with enterprise partners.

• →

  Lead the OI Team — Hire, level, and grow ML engineers and research scientists across formulation, prediction, and generative models, and set the review standards they work under.

• →

  Set Model Roadmap — Decide what we double down on, retire, or merge across brief-to-formula generation, reformulation and cost optimization, molecule generation, deformulation, odor prediction, and per-customer property models.

• →

  Own the Benchmarks — Run the OI benchmark suite, including internal quality gates, contractual acceptance criteria, and the model cards that go to customers and investors, so "is this model better?" gets a fast, credible answer.

• →

  Deliver Enterprise Commitments — Be accountable for contractual model deliverables with hard accuracy gates and quarterly dates, and say early and clearly when a target is out of reach.

• →

  Run the Data Flywheel — Work with Sensory Science and Fragrance Development to design experiments, define annotation protocols, and set the KPIs that turn panel time into model gains.

• →

  Ship to Production — Partner with Software Platform and Product Engineering to move models from research to a reliable production API, with versioning, monitoring, and drift handled.

• →

  Stay Hands-On — Read diffs, challenge eval design, and run experiments yourself.

• →

  Make Hard Cuts — Kill good work to make room for essential work, and explain the call with kindness and clarity.

• →

  Partner Across Osmo — Work daily with Chemistry, perfumers, and evaluators so model output is grounded in what people actually smell.

• →

  Own the Bar — Define what excellent applied ML looks like at Osmo and hold the team, and yourself, to it.

• Seasoned Applied ML Leader — 10+ years of industry ML/AI experience in an applied or research capacity, with models shipped to real users, not only papers or prototypes.

• Proven People Leader — You have managed, hired, leveled, and grown ML engineers or research scientists. Team lead experience with real ownership counts; we care about substance over title.

• Deep in Your Craft — Depth in at least one of: generative models over structured or discrete objects, molecular or chemical ML, multi-modal representation learning, LLM systems and evaluation, or Bayesian and experimental design. Breadth across several is a strong plus.

• Rigorous on Evaluation — You ask what the baseline is, whether the test set leaked, and whether the metric moved for the reason everyone assumes.

• Production Minded — You have owned models in production and know what versioning, monitoring, drift, and on-call actually cost.

• Curious About Hard Data — You are energized by small, noisy, expensive data and by getting the most out of hard-won human perceptual labels rather than internet-scale ones.

• Tenacious and Kind — You hold a high bar and stay with hard problems, and you lead the team with candor and care.

• Hands-On — You still read diffs and run experiments. At a company this size, a leader who can't do the work can't lead it.

• Hybrid schedule, 3 days a week in the Elizabeth, NJ office

• Science Background — Chemistry, chemoinformatics, sensory science, or biology experience.

• External Deliverables — Experience in an enterprise or contract-research setting where you owned model deliverables for outside partners.

• Built the Foundation — Experience standing up an ML platform or eval infrastructure from scratch.

• TBD, not yet confirmed by Rich. Draft based on the JD: a fast, credible, reproducible OI benchmark suite; enterprise accuracy gates hit or flagged early; a hiring bar and leveling standard in place; and clear keep, retire, or merge decisions across the six model areas. Confirm at kickoff.