Professional Experience
of Thomas P. Stockfisch

Stockfisch Consulting, Escondido .............. 2008 - present
Providing consulting services in statistical prediction model building, source-code based quality assurance, and custom algorithm development.

ACCELRYS INC, San Diego, CA ................... 1990 - 2007
Leading provider of scientific business intelligence software for the integration, mining, analysis, modeling and simulation, management, and interactive reporting of scientific data. Customers include most of the world’s leading pharmaceutical, chemical, biotechnology, nanotechnology, academic, and government research organizations.
Manager, Principal Scientist

• HIGHLIGHTS
  • awarded 3 solo algorithm patents
  • devised and implemented 17 new algorithms
  • 31 software releases involving 12 million line code base
  • parameterized 17 functional groups for CFF force field (>8,000 parameters)
  • presented 21 technical talks externally
  • managed, motivated, and mentored teams of talented scientists/developers
  • cut maintenance costs in half for large (500,000 line) code base through re-factoring
  • sped up existing Decision Tree code by factor of 200
• Software Development
  • Designed and implemented functionality for:
    • grid-like computation on large set of workstations
    • parameter determination with constraints
    • back-propagation neural networks
    • recursive partitioning
    • linear discriminant analysis
    • high-speed file-based methods for molecular properties
    • algebraic input for molecular spreadsheet system
    • reliable signaling of spawning processes
    • interactive graphical display of proteins, networks, and RP trees
    • file format inter-conversion
    • large file support (>4 gigabytes)
    • unique smiles and substructure search for zeolites
    • force field atom typing, charge library lookup, and parameter management
    • force field sampling for training set generation
  • Devised and maintained extensive test suites
  • Improved performance of scientific computation bottlenecks
  • Ported code from IRIX to LINUX and Windows XP
  • Made architectural improvements to code base:
    • merged diverging variants
    • separated scientific calculation from UI
    • added hierarchical structure to libraries
    • re-factored memory management
    • commented side effects and interfaces
    • eliminated global variables
    • improved modularity
  • Wrote tutorials and documentation
  • Evaluated potential 3rd party algorithms and software
  • Answered questions from customer support
  • Contributed to the products:
    • Discovery Studio
    • InsightII
    • Cerius2
    • TopKat
    • Medchem Explorer
    • PEFC software
  • Algorithm Development
    • Devised algorithms for:
      • force field development
      • QSAR modeling, resulting in 3 solo patents
      • substructures search
      • high performance file-based molecular property data
    • Assisted with patent applications
  • Conducted annual training of sales/support staff in new features
  • Managerial
    • performed candidate screening, interviewing, and hiring
    • conducted annual reviews
    • coordinated Cerius2 linux port involving 25 developers, 3 sites, and 5 million lines of code
    • developed and maintained 5 year plan for CFF force field development
    • estimated costs for new projects

UNIVERSITY OF CALIFORNIA, SAN DIEGO ................. 1990
Postdoctoral Research Chemist

• Applied back-propagation neural network to Monte Carlo-determined partition functions for the exact solution of the 3D Ising model on finite lattices

UNIVERSITY OF CALIFORNIA, SAN DIEGO ................ 1981 - 1989
Research/Teaching Assistant

• Developed a lattice model for microemulsions
• Calculated phase diagrams using a mean-field approximation
• Devised a novel method for obtaining model surface tension
• Used a Monte Carlo simulation to demonstrate conclusively the existence of three-phase homogeneous equilibrium and ultra-low surface tension in the model
• Developed scientific software for series analysis and general interactive numeric programming
• Worked on a method for solving the three-dimensional Ising model for non-zero magnetic field using real-space renormalization group theory
• Devised a generalized decorated lattice model for an arbitrary number of components and reactions to settle a controversy concerning the anomalous slowing of chemical reactions near a critical point
• Taught:
  • quantitative analysis lab
  • senior physical chemistry lab
  • quantum mechanics problem section
  • introductory chemistry problem section
  • substitute lecture for professor in quantum mechanics course