Applied AI Engineer

Job DescriptionAPPLIED AI ENGINEERLocation: San Francisco Bay AreaCOMPENSATIONTrial Contract: Applied AI Engineer$70–$95 per hour; approximately 20 hours per week$70,000–$100,000 annualized at that schedule.Potential Follow-On Full-Time Contract: Founding Applied AI Engineer$140,000–$200,000 annually plus 0.10%–1.00% equity, depending on experience, trial-contract performance, demonstrated technical ability, responsibilities and final role scope.Trial is expected to last six weeks. Strong performance may lead to a three-to-six-month extension and potential full-time contract.HOW TO APPLYApply through LinkedIn or email info@panoptic.bio with:Your CV or LinkedIn profile.A sample of your work: predictive model, repository, technical paper, model card or machine-learning project you personally built.A short explanation of the prediction target, validation design and most important failure mode.A short answer: How would you prevent temporal leakage when forecasting clinical-trial outcomes?Do not send confidential or proprietary material.ABOUT PANOPTIC BIOPanoptic Bio builds quantitative clinical-trial intelligence for biopharma. We combine structured scientific diligence with AI models to forecast clinical outcomes, identify failure risks and improve drug-development decisions.THE ROLEWe are hiring an Applied AI Engineer who can build rigorous predictive models for clinical-trial intelligence.You will work with small, noisy, temporal and heterogeneous datasets across clinical trials, drug development, biology and scientific literature. You must be able to define prediction targets, build strong baselines, prevent leakage and produce calibrated forecasts that can support real decisions.WHAT YOU WILL DOForecast clinical-trial duration and time to important milestones.Forecast clinical catalysts, phase transitions and trial outcomes.Identify, label and predict clinical-trial failure modes.Model biological, clinical, enrollment, regulatory and manufacturing risks.Build intelligence around failed assets and potential asset-revival strategies.Develop strong baselines using classical statistical and machine-learning methods.Build and evaluate XGBoost, LightGBM, CatBoost and deep-learning models.Evaluate tabular foundation models such as TabPFN.Evaluate biological foundation models where they improve prediction or representation quality.Combine structured trial data, scientific literature, regulatory information and biological data.Design leakage-resistant, time-aware training and evaluation frameworks.Produce calibrated probabilities, uncertainty estimates and detailed error analyses.Build reproducible data, training, evaluation and inference pipelines.Translate model outputs into useful clinical-trial intelligence.Work closely with Panoptic’s clinical-development scientists.WHO WE ARE LOOKING FORStrong Python and applied machine-learning skills.Experience with scikit-learn, XGBoost, LightGBM or CatBoost.Experience with PyTorch, JAX, TensorFlow or an equivalent deep-learning framework.Strong understanding of tabular modeling, feature engineering and model evaluation.Understanding of calibration, uncertainty estimation, class imbalance and distribution shift.Familiarity with survival analysis, censored outcomes or time-to-event modeling.Ability to design temporal splits and recognize target or feature leakage.Evidence that you have personally built, evaluated and shipped predictive models.Ability to compare simple and complex methods objectively.Ability to communicate model assumptions, limitations and trust boundaries clearly.High ownership, speed and intellectual honesty.Experience with biotechnology, clinical development, healthcare data, scientific machine learning or biological foundation models is valuable but not required if you can learn the domain quickly.A graduate degree in computer science, machine learning, statistics, applied mathematics, computational biology or a related field is preferred. Exceptional evidence of applied modeling ability matters more than the exact credential.REQUIRED READING AND APPLICATION CHECKTo show that you understand the job description and are not automating your outreach, please use “clinical trial intelligence” as the first three words of your application or outreach message.Please read the following articles and review the biological foundation-model registry. Be prepared to answer questions about them during the interview process.Trial Terminal Biological Foundation-Model Registry:https://trialterminal.bio/lab?view=modelsPanoptic Bio’s Clinical Trial Outcome Prediction:https://www.clinicaltrials.bio/p/panoptic-bios-clinical-trial-outcomehttps://www.clinicaltrials.bio/p/panoptic-bios-clinical-trial-outcome-425Trialbench - which has datasets and tasks that form a good introhttps://www.nature.com/articles/s41597-025-05680-8IMPORTANT DISCLAIMERParticipation in recruiting, interviews, assessments or related discussions does not guarantee selection for or acceptance into a trial contract. An offer exists only when confirmed in a written agreement signed by Panoptic Bio and the candidate.Participation in or completion of a trial contract does not guarantee an extension, full-time contract, employment, equity, benefits or any other compensation or entitlement. Any potential extension, full-time arrangement, equity grant, title, compensation or benefit remains subject to performance, business needs, applicable approvals, applicable law and a separate written agreement.