Basis Postdoctoral Fellow — LLM-Based Reasoning and Planning
About the Fellowship
This Basis Postdoctoral Fellowship is a collaborative initiative between Basis Research Institute and Harvard University’s Amin Lab. It also includes collaborators at the Hugh Kaul Precision Medicine Institute. As a fellow, you will be a key contributor to our focused efforts to research and develop fundamental methods in AI to help practitioners, including doctors and civic policymakers, make informed decisions based on vast amounts of data and information.
About Basis
Basis (https://www.basis.ai/) is a nonprofit applied AI research organization with two mutually reinforcing goals.
The first is to understand and build intelligence. This means to establish the mathematical principles of what it means to reason, to learn, to make decisions, to understand, and to explain; and to construct software that implements these principles.
The second is to advance society’s ability to solve intractable problems. This means expanding the scale, complexity, and breadth of problems that we can solve today, and even more importantly, accelerating our ability to solve problems in the future.
To achieve these goals, we’re building both a new technological foundation that draws inspiration from how humans reason, and a new kind of collaborative organization that puts human values first.
About Nada Amin’s Group
Nada Amin’s research group explores new ways of programming that are easier, safer, faster. For safer, we look at systems and formalisms based on types and verification. For faster, we look at meta-programming techniques, including generative programming and reflection, to collapse levels of interpretations and move between different views of the same program in a way that helps optimizations, understanding, and modifications. For easier, our goal is to enable more people to manipulate computer programs (static) and processes (dynamic) robustly. To this end, we look at combining Machine Learning and Programming Languages to enable the creation of neuro-symbolic systems that can move back and forth between learnable (neural) and interpretable (symbolic) representations of a system.
Research Focus
Our research aims to develop new foundations and technologies for reasoning and planning that leverage large language models (LLMs) and integrate them with complementary approaches, including program synthesis, and causal and probabilistic programming. The goal is to create integrated systems that harness the deep domain knowledge and emergent capabilities of LLMs, alongside the precision, efficiency and correctness of formal reasoning frameworks. This could deliver the best of both worlds: the intuitive, context-rich insights of LLMs and the logical, structured analysis provided by formal methods. Key application areas include developing integrated agents that:
- Automate data science: augmenting and automating functionalities in the data science pipeline, such as data curation, cleaning, format translation, causal and statistical analysis, and model verification.
- Reliably execute long-term plans using heterogeneous external tools: improving automated planning and decision-making in complex open environments.
Fellows will have the opportunity to develop these tools within concrete projects with real-world impact, leveraging the collective networks of Basis, Harvard, and Hugh Kaul Precision Medicine Institute. Specifically, fellows will be able to contribute to larger initiatives to (i) dramatically advance precision medicine, or (ii) advance society’s capacity to make informed civic policy decisions, building upon Basis work in civic policymaking.
The research environment is both structured and adaptable, with multiple avenues for scholarly contribution. As a fellow, your expertise can shape various aspects of the project. It will deepen your foundations in multiple domains, including large language models, reasoning, planning, and precision medicine or civic policymaking while also allowing for a balance of focused research, academic exploration, and software development.
Who we’re looking for
- Researchers holding a PhD related to programming languages, artificial intelligence, and machine learning. Researchers with experience in other adjacent technical areas such as physics or mathematics will also be considered.
- Experience in machine learning, particularly with (i) language models and (ii) reinforcement learning and planning, is highly valued.
- Interest in end-user applications, especially in precision medicine or policymaking.
- Individuals with a demonstrated track record in scientific research, which can be evidenced through publications, technical reports, or impactful software projects.
Core Responsibilities
- Conduct independent and collaborative research, focusing on reasoning and language models
- Apply these methods to concrete applications within precision medicine or civic policymaking
- Disseminate research findings through academic publications and presentations at leading conferences.
- Actively engage in knowledge transfer within Basis and Harvard, converting research into actionable insights and algorithms.
- Provide mentorship to junior team members and contribute to the scientific discourse through seminars, workshops, and collaborative projects.
Role Details
- Full-time: This fellowship is full-time and has a fixed duration of 1 to 2 years.
- Location: Applicants must reside in, or be willing to relocate to, the Greater Boston area. This is an in-person position — you will have space at both Nada Amin’s lab and at Basis. You will be expected to travel periodically, about once every six to eight weeks, for Basis-wide in-person events, typically in New York.
- Salary: Competitive with leading postdoctoral fellowships.
- Start date: The fellowship will begin in early spring to summer, with the exact date to be determined.