Senior / Principal Scientist, Molecular Simulations

Flagship Pioneering•Cambridge, MA

About The Position

FL117, a venture-backed stealth AI x bio company, is seeking a Sr. Scientist / Principal Scientist, Molecular Simulations. Here you will build the physics layer of a drug development AI platform. You will characterize how binders engage their targets and validate generated designs before they are committed to synthesis. You will work at the interface of structure-based modeling and generative ML, turning simulation output into training signals. We are looking for a hands-on simulation scientist who is comfortable owning a method end to end in a cross-functional, fast-moving team.

Requirements

PhD in computational chemistry, biophysics, chemical physics, structural biology, or a related field; MSc with equivalent industry depth considered
4+ years of post-PhD molecular simulation experience, at least part of it in a drug discovery setting (Sr. Scientist); 6+ years with demonstrated ownership of a method or platform (Principal)
Deep hands-on expertise in classical molecular dynamics: force fields and small molecule parameterization (AMBER/CHARMM/OPLS, GAFF/OpenFF), solvation and ion treatment, equilibration protocols, and the characteristic failure modes of each.
Production experience with at least one enhanced sampling or free energy framework (relative or absolute FEP/TI, metadynamics, umbrella sampling, replica exchange, or weighted-ensemble MD) including validation against measured affinities.
Structure-based drug design fundamentals: docking and pose evaluation, pharmacophore and hot-spot analysis, and critical assessment of experimental and predicted structures.
Strong Python, with fluency in simulation toolchains (OpenMM, GROMACS, AMBER, NAMD, or Desmond) and analysis stacks (MDAnalysis/MDTraj, RDKit); version control and reproducible workflows as habit, not aspiration.
Demonstrated ability to run simulations at scale on GPU, HPC, or cloud infrastructure through automation rather than manual per-system setup.
Judgment about method cost versus decision value: able to say when a multi-week free energy campaign is warranted and when docking plus a short MD run is enough.
Consistent record of outstanding technical output reflected in publications, patents, or high impact internal reports

Nice To Haves

Peptide, macrocycle, or other beyond-Rule-of-5 simulation: conformational ensembles, intramolecular hydrogen bonding, chameleonicity, and membrane permeability (ex: water-to-membrane transfer free energies).
Experienced with target classes such as: GPCRs, protein–protein interactions, transcription factors, highly dynamic target classes.
Working with predicted structures and their limitations (AlphaFold-class models, co-folding methods), and with cryptic or induced-fit pocket discovery.
ML-adjacent simulation work: machine-learned force fields and potentials (ANI, MACE, NequIP-class), learned scoring functions, or generating simulation-derived features for downstream generative models.
Workflow and infrastructure tooling: Snakemake/Nextflow, Airflow/Prefect/Dagster, Docker, AWS (Batch, ParallelCluster, S3), and experiment tracking (MLflow/W&B).
Clear structural communication to non-specialists using PyMOL, ChimeraX, or Mol.
Strong presence in the computational chemistry or biophysics community through publications and conference proceedings.

Responsibilities

Run and interpret structural and dynamic analysis of ligand–target complexes peptides, macrocycles, nanobodies/VHHs, antibody CDRs.
Characterize binding sites, including cryptic and induced-fit pockets, across the target portfolio using MD, enhanced sampling, and pocket detection methods.
Triage generated molecules before synthesis through a combination of pose prediction and refinement, docking rescoring, binding-mode stability under MD, ligand strain and conformational ensemble analysis, and define the criteria by which a design advances or gets killed.
Build, validate, and own the affinity prediction workflow (relative and absolute FEP, MM-GBSA/PBSA, endpoint methods); establish its accuracy and applicability domain against measured data and report both honestly.
Model conformational behavior, intramolecular hydrogen bonding, chameleonicity, and membrane partitioning for macrocycles and other non-classical chemotypes to support oral property design.
Convert simulation output into structured signal for the generative platform in close partnership with the AI/ML team.
Build reproducible, automated simulation pipelines on GPU cloud or HPC infrastructure; own system setup, parameterization, QC, and analysis at portfolio scale rather than one system at a time.
Work directly with medicinal chemistry, pharmacology, and external structural biology and assay partners; turn structural hypotheses into testable designs, feed returning data back into corrected models, and flag clearly where predictions are unreliable.