Alexander Feldman

Staff Embedded Systems Engineer — Atomic Machines

Feb 2025 – Mar 2026 · Emeryville, CA

• Worked on distributed control architecture for modular manufacturing robots composed of precision networked nodes, each built around an AMD Xilinx KR260 SoM with custom serializer/deserializer hardware communicating with STM32 motor-control boards.
• Identified a critical networking bottleneck on the Cortex-R5 firmware path where a single polling iteration over all four PMOD channels consumed 800 µs against a 1 ms target.
• Designed and integrated DMA IP into the Vivado FPGA bitstream, reducing the full polling iteration to less than 800 µs s—a 40× improvement—eliminating the bottleneck and enabling additional network scaling, more stations, and higher effective bandwidth.
• Tuned PID control loops for multiple linear stages and motorized axes using register-level configuration of TMC motor-control ICs.
• Took end-to-end ownership of a bonding-press node: firmware integration with the F-Link TMC platform, electrical assembly of pre-made PCBs, motor installation, process-control software, and a Flet-based Python user interface.
• Implemented a Python DSL for bonding recipes and achieved stable multi-hour production runs with temperature control converging to 2 °C and force control to 5 N.

Senior Electrical Engineer — NextFlex

Nov 2022 – Feb 2025 · San Jose, CA

• Completed a complex Siemens Expedition PCB design for a Ka-band radar front-end in collaboration with Lockheed Martin, continuing DARPA IceCold work on jet-impingement cooling for high-power electronics.
• Inherited a design with numerous DRC violations, unconnected nets, and manufacturability problems in a compact RF layout containing waveguide structures and coolant cavities; resolved all issues and brought the design to fabrication readiness.
• Identified and engaged a fabrication partner willing to prototype the board after multiple PCB manufacturers declined the project due to its complexity.
• Performed microscope-assisted post-processing and electrical testing of fabricated hardware; investigated leakage caused by substrate material limitations.
• Redesigned and implemented the complete front-end software stack for a textile touch-pad interface used in wearable soldier computing systems developed with AFFOA and Penn State.
• Ported the system to the Nordic nRF52840, integrated TensorFlow Lite for embedded gesture recognition, implemented Bluetooth communication, and developed host-side software for training and evaluation.
• Introduced Zephyr RTOS into the organization and became the primary internal resource for nRF52840 and Zephyr-based development.
• Participated in demonstration days and technical outreach for visitors and educational programs focused on hybrid electronics.

Member of Research Staff — Xerox PARC

Sep 2014 – Sep 2022 · Palo Alto, CA

• Conducted research at the intersection of artificial intelligence, algorithm design, diagnostics, electronic design automation, and cyber-physical systems within the System Sciences Laboratory, Model-Based Reasoning Area.
• Designed and built an instrumented refrigerator testbed to compare machine-learning and model-based diagnosis on real physical systems, including custom data-acquisition electronics (dual-channel ADC PCB), failure-injection hardware (relay-controlled compressor, linear actuator for door), embedded firmware (ChibiOS on Arduino-class hardware), and a Python Flask experiment-scheduling system collecting data at ≈ 60 Hz over 24-hour sessions.
• Trained and evaluated neural networks and decision trees against model-based diagnosis algorithms on collected sensor data (temperature, voltage, current), producing comparative results across methods.
• Designed the first general class of algorithms for multi-output Boolean circuit synthesis that are sound, complete, and optimal, using SAT and quantified Boolean formula (2-QBF) encodings; connected results to realistic VHDL/Verilog design flows and component-library-based synthesis.
• Developed quantum formulations of diagnostic problems for D-Wave-style quantum annealers; this work was cited in a DARPA BAA calling for quantum approaches to diagnosis.
• Partnered with Northrop Grumman and three other companies on quantum computing proposals for defense applications.
• Designed and implemented a Condition Based Maintenance (CBM) platform providing diagnostic and prognostic analytics, sensor-placement optimization, SCADA interfacing, and code generation for diagnostics in control systems.
• Built a SPICE model of a Field-Programmable Analog Array (FPAA) for analog computing research, leading to a DARPA RFI response.
• Co-developed the DXC diagnostic algorithm evaluation framework jointly with NASA Ames; organized three International Diagnostic Competitions (DXC'09, DXC'10, DXC'11).
• Built substantial research software in C/C++ and Python for diagnosis, benchmarking, synthesis, and experimentation; selected systems are being prepared for open-source release.

Technical Consultant II — Nspyre B.V./ASML

Jul 2013 – Aug 2014 · Eindhoven, The Netherlands

• Technical consulting on advanced engineering systems at ASML, the world's sole supplier of EUV lithography machines for semiconductor manufacturing.

Founder and President — General Diagnostics

Jun 2013 – Sep 2014 · Delft, The Netherlands

• Founded a company to commercialize model-based diagnosis technology developed during doctoral and postdoctoral research.

Postdoctoral Research Fellow — University College Cork

Jun 2012 – Jun 2013 · Cork, Ireland

• Postdoctoral research in the Complex Systems Laboratory on model-based diagnosis of analogue and hybrid systems.

Postdoc — HEIG-VD & EPFL (visiting)

Jun 2010 – Apr 2012 · Switzerland

• Hired by Prof. Stephan Robert at the Institute of Information and Communication Technologies (IICT), HEIG-VD, with concurrent visiting positions at EPFL's RF Integrated Circuit Group (Prof. Catherine Dehollain) and Distributed Intelligent Systems and Algorithms Laboratory, DISAL (Prof. Alcherio Martinoli).
• Worked on ultra-wideband (UWB) indoor localization for mobile robots as part of the Swiss NCCR MICS initiative funded by the Swiss National Science Foundation, in collaboration between HEIG-VD and two EPFL labs.
• Took a bench-top single-transmitter/receiver RF experiment and turned it into a fully working real-time multi-receiver robotic localization system with cm-level precision using Impulse-Radio UWB at 4.25 GHz center frequency with 500 MHz bandwidth.
• Designed and implemented the complete FPGA-based digital back-end for four wall-mounted UWB receiver base stations processing aggregate pulse trains with cryptographic time-hopping sequences and computing Time Difference of Arrival (TDOA) ranging.
• Overcame a timing-closure failure on the FPGA by overclocking it via speed-grade faking in the synthesis tools—a calculated risk that worked and brought the entire system online without months of hardware delay.
• Developed a Qt-based real-time graphical user interface displaying live robot trajectories as Khepera III platforms drove around the arena.
• Co-authored three peer-reviewed publications: IEEE VTC'11-Fall, PRIME'11 (Gold Leaf Certificate), and IEEE ICUWB'12.
• Collaborated with Dr. Alexander Bahr (DISAL/EPFL) on numerical simulation models for UWB localization design-trade-off analysis (accuracy vs. update rate, receiver placement).
• Continued diagnostic reasoning research during this period, presenting work on stochastic search methods at DX'11 in Munich.
• Independently identified and won a competitive ESA contract—the GENIUS project (Ground-basEd diagNostIc sUpport System)—funded under ESA's Innovation Triangle Initiative (ITT AO/1-6067/09/NL/CBI), total value 150000 €, executed concurrently with the full-time UWB postdoc.
• Assembled a cross-institutional consortium: inventor (TU Delft IP), developer (Logica Nederland B.V., later CGI), and end user (ESA-ESOC); subcontracted through Logica with Helena Vicente de Castro as technical contact.
• Built a ≈1000-line LYDIA-NG model of the GOCE satellite Electrical Power System (1397 variables: 186 observable, 289 health, 132 command, 790 internal) and implemented the complete GENIUS framework: diagnostic search (BFS, DFS, BGSS), SPICE-based simulation with constraint propagation, residual analysis, entropy-based active testing (FRACTAL), and an API to ESA's SCOS-2000 SCADA and SIMSAT/PSS infrastructure.
• Designed 11 fault scenarios (single through quadruple faults, ambiguous, intermittent, degradation) with independent telemetry from the ESA-ESOC SIMSAT simulator. Achieved diagnostic accuracy of 1.0, average classification error of 1.25 (out of 289 components), fault detection in < 100 ms, and diagnostic uncertainty reducible by a single automatically computed telecommand.
• First-ever application of a model-based active testing algorithm to a real-world space system; published at IEEE Aerospace Conference 2013 (14-page paper) and DX'13.

Ph.D. Student / Doctoral Research Fellow — Delft University of Technology

Sep 2005 – Sep 2009 · Delft, The Netherlands

• Conducted doctoral research on approximation algorithms for model-based diagnosis, active testing, and stochastic search under Prof. Arjan van Gemund.
• Published results at IJCAI, AAAI, and in the Journal of Artificial Intelligence Research (JAIR).
• Visiting researcher at Xerox PARC (May–Sep 2008).

Software Architect — Science & Technology B.V.

Apr 2005 – Sep 2005 · Delft, The Netherlands

• Software architecture for scientific and engineering systems.

Senior Programmer / Market Database Controller — ING Bank

Dec 2002 – Apr 2005 · Amsterdam, The Netherlands

• Worked in market risk management on data-intensive banking systems and market databases.

Senior Programmer — Zend Technologies

Jul 2000 – Sep 2001 · Ramat Gan, Israel

• Early-career software development at the company behind the PHP language engine.

Ph.D. in Computer Science (cum laude) — Delft University of Technology

2010-05-17 · Delft, The Netherlands

Thesis: Approximation Algorithms for Model-Based Diagnosis. Advisor: Prof. Arjan van Gemund.

M.Sc. in Computer Science, Technical Informatics (cum laude) — Delft University of Technology

2005-03-29 · Delft, The Netherlands

Thesis: Hierarchical Approach to Fault Diagnosis. Advisor: Prof. Arjan van Gemund.

B.Sc. in Computer Science — University of Economics – Varna

2000-06-30 · Varna, Bulgaria

Secondary vocational diploma — Vocational High School of Economics “Georgi Sava Rakovski”

1995-06-30 · Varna, Bulgaria