Bimal Gaudel

C++ Systems Engineer · Numerical Libraries · Simulation & Scientific Computing

Open to work. Looking for C++ systems and runtime roles, numerical and simulation libraries, scientific software R&D, and HPC research-software / postdoc roles. Based in Blacksburg, VA. Open to relocation across the US, Canada, and Europe (priority in that order). Visa sponsorship required.

bimalgaudel@gmail.com Download CV

C++17/20/23

I build production C++ systems and numerical libraries for scientific computing — the runtimes, solvers, and library cores other engineers and researchers ship on top of. Not a theorist who codes; an engineer who understands the science.

Technical Expertise

• Modern C++ (17/20/23). Production-grade runtimes, template metaprogramming, and extensible APIs. Designed SeQuant’s backend interface so cleanly that an external contributor integrated a new numerical framework in under 30 minutes.
• Numerical & simulation libraries. Block-sparse distributed tensor framework (TiledArray) extended with tensor-of-tensor data structures for PNO-like local-correlation methods; DP-based contraction ordering using runtime index extents; graph-theoretic optimization (TNCO, CSE).
• HPC & distributed runtimes. MPI distributed-memory programming, memoization-cache design, and supercomputer-scale execution validated in MPQC production on NERSC Perlmutter.

SeQuant — Lead Designer of a Complete Compiler Pipeline

I designed and implemented SeQuant from scratch over seven years — embedded DSL, graph-canonicalized IR, DP optimizer, tree-walk interpreter, and MPI runtime. The system automates the path from symbolic quantum-chemistry derivations to supercomputer-scale numerical execution, compressing months of researcher development work into days.

• Embedded DSL + canonicalized IR. SeQuant expressions serve directly as the AST in idiomatic C++; the IR is a full binary tree with canonical node identity from tensor-network graph canonicalization, enabling robust common-subexpression elimination across separate equations.
• Runtime DP optimizer + tree-walk interpreter. Bottom-up dynamic-programming contraction ordering using actual index-space extents; post-order traversal with canonical-identity-keyed memoization and pending-use cache eviction. Validated in MPQC production on NERSC Perlmutter.
• Extensible backend. Interface clean enough that an external contributor integrated a new numerical framework in under 30 minutes; BTAS and TiledArray backends ship with SeQuant. IR also lowers to C++/Python targeting downstream tensor compilers (TACO).

Full architecture described in (Gaudel et al., 2026).

Defended December 2025. Slides.

TiledArray

Contributing developer on TiledArray, the massively-parallel block-sparse distributed tensor framework that backs SeQuant’s runtime. Extended its operation basis to support tensor-of-tensor data structures (outer-only, inner-only, and mixed contractions; TT×T and TT×TT products; full TT dot products), enabling block-wise compression and PNO-like local-correlation methods.

Experience

Research Assistant | Virginia Tech 2018 - 2025
Designed and implemented SeQuant’s compiler and runtime; contributed to TiledArray.

Education

PhD in Theoretical & Computational Chemistry | Virginia Tech 2018 - 2025
Dissertation: Automated implementation of advanced electronic structure methods

Master of Science in Physical Chemistry | Tribhuvan University 2014 - 2016

Publications

SeQuant (Gaudel et al., 2026)

A color-graph approach to canonicalizing tensor networks, used for symbolic transformation and runtime evaluation of many-body methods.

Applied research using SeQuant

• Theoretical exploration of new ansatze in explicitly correlated methods (Masteran et al., 2025).
• Geminal parameter tuning in explicitly correlated methods (Powell et al., 2025).
• Discovery of effective theories compared to complex counterparts (Teke et al., 2024).
• Identification and correction of errors in previously published works (Masteran et al., 2023).

Featured writing

Fast RTTI in C++ for a class hierarchy

Compile-time type-id generation for performant runtime type inference across a class hierarchy.

The access-by idiom in C++

A pattern for unit-testing private methods when no better seam exists.

See all posts →

References

Gaudel, B., Adam, R. G., Melekamburath, A., Masteran, C., Teke, N., Besharatnik, A., Köhn, A., & Valeev, E. F. (2026). SeQuant framework for symbolic and numerical tensor algebra. I. Core capabilities. Journal of Chemical Physics, 164(14), 142502. https://doi.org/10.1063/5.0311913

Masteran, C., Gaudel, B., & Valeev, E. F. (2025). Toward a balanced description of ground and excited states with transcorrelated F12 methods. Journal of Chemical Theory and Computation, 21(20), 10329–10339. https://doi.org/10.1021/acs.jctc.5c01434

Masteran, C., Kumar, A., Teke, N., Gaudel, B., Yanai, T., & Valeev, E. F. (2023). Comment on “canonical transcorrelated theory with projected slater-type geminals” [J. Chem. Phys. 136, 084107 (2012)]. Journal of Chemical Physics, 158(5), 57101. https://doi.org/10.1063/5.0135257

Powell, S. R., Surjuse, K. A., Gaudel, B., & Valeev, E. F. (2025). Slimmer geminals for accurate F12 electronic structure models. Journal of Chemical Theory and Computation, 21(18), 8833–8842. https://doi.org/10.1021/acs.jctc.5c00971

Teke, N. K., Melekamburath, A., Gaudel, B., & Valeev, E. F. (2024). “Best” iterative coupled-cluster triples model? More evidence for 3CC. Journal of Physical Chemistry A, 128(45), 9819–9828. https://doi.org/10.1021/acs.jpca.4c04667