Machine Learning/AI Scientist for Data Science Team

ANDREW LAWRENCE STACHYRA

ADDRESS:

** **** **.

Arlington, MA 02474-2816

781-***-****

***************@*****.***

SUMMARY:

Technically oriented professional with physics Ph.D. in search of challenging assignment in product research, design, development, analysis, or testing, using established core competencies:

• Statistical data analysis

• Data visualization

• Algorithm design and evaluation

• Machine learning

• Modeling and simulation

• Scientific computing

• Real-time computing

• Signal and image processing

• Debugging and troubleshooting

• Technical writing and presentation

EXPERIENCE:

Principal Research Scientist, Physical Sciences Inc., Andover, MA

July 2011-January 2013

• Developed unsupervised machine learning technique using expectation maximization algorithm to estimate track parameters for multi-object tracking scenarios with an ambiguous number of underlying objects

• Implemented image and signal processing algorithms such as Hough transform and order statistics constant false alarm rate (CFAR) threshold, to investigate use in mitigating sensor saturation effects due to radar clutter

• Conducted time series analysis of radar and infrared signature data from ballistic missile objects tracked during Aegis BMD FTM-16 Event 2 flight test mission, to demonstrate potential utility in augmenting sensor-to-sensor track correlation algorithms which currently use only object position data

• Built software interface to enable integration of third party-supplied infrared thermal emissivity modeling support into MATLAB/Simulink-based missile object tracking simulation tool

• Took lead role in numerous customer facing interactions, including delivery of multiple viewgraph presentations as well as authoring white papers, contract deliverable technical progress reports, and a software user manual

Systems Engineer, Raytheon Integrated Defense Systems, Woburn, MA

November 2006-April 2011

• Participated in year-long program to develop supervised machine learning and statistical classification techniques for predicting the ide ntities of unknown missile objects based on features observed in their radar signatures

• Conducted basic research into potential use of object time series signature data or “target fingerprint” to augment performance of radar-to-radar and radar-to-infrared track correlation algorithms

• Coordinated internal vetting, development, editing, ranking and assembly of 38 missile defense research proposal suggestions, submitted by 15-20 colleagues across multiple time zones and business units, to create $5 million contract renewal package for final delivery to government project office

• Lead four person analysis team in risk reduction effort, prior to flight test mission, to review large number of scenario simulations in search of sporadic or difficult-to-repeat errors in X-band radar tactical control software

• Performed root cause analysis to troubleshoot and propose corrective action for anomalies in UHF-band missile defense radar tactical control software

Staff Member, MIT Lincoln Laboratory, Lexington, MA

September 2002-October 2006

• Provided missile defense systems engineering consulting services to government sponsor

• Implemented MATLAB realizations of sensor-to-sensor track correlation, multi-sensor data fusion, coordinate transformations, Kalman filtering, and various missile defense system command and control functions, for broad use in conducting missile defense performance studies

• Designed specialized data compression technique for condensing radar track information, to reduce communications message bandwidth by leveraging similarities between multiple adjacent tracks in a threat missile complex

• Conducted end-to-end system vulnerability study to assess effects on missile defense system performance when sensors report statistically inaccurate estimates for track uncertainty or the track parameter covariance matrix

Research Assistant, University of Washington Physics Department, Seattle, WA

June 1994-February 2002

• Wrote grant proposal resulting in $18,600 student research award to participate in construction of large scale underground particle physics detector in Japan

• Designed custom-built real-time C language TCP/IP sockets software to allow integration, via network connection, of data from one detector subcomponent into main data stream for entire detector experiment

• Developed and applied innovative tomography-like imaging technique to create low-resolution topographical map of mountain surface above underground particle detector, based on penetration of cosmic rays through rock overburden

• Manipulated and data-mined extremely large (TByte-sized) particle physics data set, then performed statistical analyses on smaller, abstracted data sets in search of new and fundamental physics results

EDUCATION:

Ph.D., Physics, March 2002, University of Washington, Seattle, WA

Thesis Title: “A Search for Astrophysical Point Sources of Neutrinos with Super-Kamiokande”

M.S., Physics, December 1994, University of Washington, Seattle, WA

B.S., Physics, May 1993, Yale University, New Haven, CT

SKILLS:

Experienced with MATLAB, UNIX/Linux, C, legacy Fortran, Word, Excel. and

PowerPoint. Reference-assisted use of version control tools (CVS, Subversion, and

ClearCase) and scripting languages (e.g., /bin/sh, /bin/awk, etc.).

PROJECTS:

http://www.mathworks.com/matlabcentral/fileexchange/39872-expectation-maximization-1d-tracking-demo

Contact this candidate