About the RoleTBC is seeking a Computational Neuroscientist to help derive novel algorithms and model improvements for AI from understanding the dynamics of real neurons.
You will work across computational neuroscience, biology and machine learning to design experiments, analyze large-scale neural recordings and build models that connect living neural systems with modern foundation models. Your work will sit at the center of TBC's Algorithm Discovery Platform: identifying where AI models fail, studying how biological neural networks approach related problems and translating what we learn into usable software.
This is a hands-on, high-ownership role for someone who wants to help define a new field. You will work closely with wet-lab biologists, AI researchers and engineers to move from experiment to mathematical principle to model performance.
Design biological computing experiments- Design experiments that encode temporal, spatial and multimodal information into living neural cultures.
- Develop stimulation and information-encoding paradigms for high-density multi-electrode array systems.
- Define experimental controls, baselines and validation criteria that distinguish useful biological effects from noise or generic dynamical behavior.
- Partner with the biology team to improve culture readiness, experimental consistency and reproducibility.
Analyze neural population dynamics- Analyze large-scale electrophysiological recordings from high-density MEAs and related neural-interface platforms.
- Model neural population dynamics, latent spaces, neural manifolds, temporal structure, effective connectivity and state transitions.
- Develop methods for decoding neural responses and identifying computationally useful spatial and temporal patterns.
- Characterize how neural networks respond, adapt, learn and retain information across different stimulation conditions and time scales.
Translate biology into AI systems- Work with AI researchers to convert neural dynamics into mathematical principles, architectures, adapters, optimizers and learning rules.
- Test whether biologically derived principles improve generative video, world models, inference efficiency, continual learning, memory or generalization.
- Compare biological approaches against strong non-biological controls and surrogate models.
- Determine which properties of the biological response are necessary for model improvement and which can be simplified for scalable software implementation.
- Evaluate discoveries across model sizes, datasets, architectures and modalities.
Build closed-loop research infrastructure- Help build tools for neural stimulation, real-time readout, experiment orchestration, data analysis and rapid iteration.
- Develop reusable analysis pipelines and computational tools that connect wet-lab experiments with AI-model evaluation.
- Support closed-loop systems in which model results inform biological experiments and biological measurements inform the next model iteration.
- Contribute to TBC's longer-term work in latent-space interfacing, neural controllability, connectome-guided learning and real-time biological inference.
Shape research strategy- Own research workstreams from hypothesis and experimental design through analysis, validation and technical communication.
- Help define research priorities, technical milestones and decision criteria for TBC's neuroscience programs.
- Identify scientific, statistical and experimental risks before they become blockers.
- Communicate findings clearly to biology, AI, engineering, product and company leadership.
- Contribute to internal documentation, research publications, technical presentations and external scientific communications as appropriate.
What Success Looks Like- Neural experiments produce consistent, high-quality and interpretable population-level data.
- Biological observations are converted into testable computational hypotheses.
- Validated neural principles become software that produces measurable improvements in real AI models.
- Results hold up against strong controls, ablations and non-biological alternatives.
- Experimental and computational pipelines allow the team to move more quickly from question to evidence.
- TBC develops a clearer understanding of how biological networks represent, transform, learn and retain information.
- Your work advances both near-term neurally optimized software and the longer-term path to real-time biological compute.
Required Qualifications- Ph.D. or equivalent research experience in computational or systems neuroscience, neural engineering, machine learning, applied mathematics, physics, statistics or a related field.
- Strong background in neural-data analysis, neural population dynamics, neural coding or dynamical systems.
- Experience working with electrophysiology, MEA recordings, calcium imaging, brain-computer interfaces or comparable neural datasets.
- Strong programming ability in Python and experience with scientific-computing and machine-learning tools.
Experience with several of the following:- Dimensionality reduction
- Latent-variable models
- Neural manifolds
- Dynamical-systems modeling
- Encoding and decoding models
- Time-series analysis
- Effective-connectivity analysis
- Statistical modeling and uncertainty analysis
- Ability to design rigorous experiments and distinguish correlation from causal or mechanistic evidence.
- Ability to communicate clearly and work effectively with wet-lab scientists, AI researchers and engineers.
- Strong scientific judgment, ownership and comfort operating in a fast-moving research environment where the playbook is still being written.
Preferred Qualifications- Experience with closed-loop neural interfaces, adaptive stimulation or real-time neural decoding.
- Experience with causal inference, connectomics, synaptic plasticity, STDP or effective-connectivity estimation.
- Familiarity with foundation models, generative video, world models, reinforcement learning or model-representation analysis.
- Experience with reservoir computing, neuromorphic computing, biological computing or other nontraditional compute substrates.
- Experience connecting population-level neural dynamics to machine-learning architectures.
- Familiarity with PyTorch, JAX or other modern deep-learning frameworks.
- Experience building reusable research infrastructure, analysis pipelines or internal scientific tools.
- Publications at leading neuroscience, neural-engineering or machine-learning venues.
- Interest in translating frontier research into products that improve real AI systems.
