Blake Anthony Wilson
Co-founder & interim CEO, Zephram · Santa Barbara, CA
Senior ML Scientist | 13+ years software & algorithm design. Specialist in designing RL and transformer infrastructure for deep-tech applications. Proven track record in optimizing data collection and deployment pipelines. Currently, co-founder and interim CEO of Zephram ($20M valuation), where I lead the development of multi-agent inference systems for frontier scientific research. PBJ — Visual multi-agent orchestration & fine-tuning
Skills
C/C++/CUDA · Python/Julia · FastAPI/PyTorch/Lightning/modal · Next.js/React/Astro · AWS/Vercel · RL/PPO/GRPO · Inference/Model Serving · ML for Physics & Optimization · Public Speaking · Technical/Grant Writing · Engineering Leadership and Management
Education
-
Ph.D. in Electrical and Computer Engineering
Purdue University
Research in machine learning for nanophotonics, device design, and optimization (NanoML / QSC). Dissertation: Adversarial Markov Processes for Quantum Generative Machine Learning and Classical Optimization
2019–2024
-
B.S. in Electrical and Computer Engineering
Purdue University
Minor in Philosophy. Optimal control for multi-agent systems and SoCET team lead.
2015–2019
Experience
Co-Founder & Interim CEO (Former CTO)
Zephram
Oxford, UK
2025 -
- Led a team of 8 Oxford engineers and scientists to develop AI infrastructure with modal, Lightning, FastAPI, PyTorch, and Next.js for building pbj, our full-stack multi-agent finetuning+inference no-code platform. Developed multi-GPU CUDA kernels for energy-model inference, achieving up to 600x speed-up over previous state-of-the-art. [1]
- Appointed by the board (Abstract) to lead our $20M company during a $1.2M financial investigation and restructuring following insurmountable fiduciary breaches from previous leadership. Negotiated partnerships with hlevel.ai, Cambridge, and UC Berkeley. Managed legal counsel and stakeholder communications to stabilize the entity. Initiated rebranding, stabilized operations, and pivoted to a new business model and product strategy.
- Proprietary Breakthroughs built on our stack: Context compaction (5x speed-up + 80% cost reduction), simulated finetuning ($0.01 vs $1000x LORA), multi-agent swarm communication protocols (15% faster on small-medium scale models), CUDA kernels for multi-GPU optimization (300x speed-up over previous state-of-the-art), and more.
Research Scientist III (AI for Quantum and NLP)
Quantinuum
Quantinuum, Oxford & London
2024 - 2025
- Led the development of internal PyTorch and CUDA packages for RL+model infrastructure, custom transformer architectures (GPT-2 and decoder heads from scratch), and dataset generation pipelines (AWS, Lightning) to support collaborations with Nvidia + Deepmind for using GPT-2 models and agentic AI to generate quantum chemistry circuits and datasets. Fueled publications and industry partnerships. [1]
- Engineered new multi-GPU CUDA kernels for UCCSD quantum circuits, achieving a 20x speed-up over previous state-of-the-art methods for data collection. Built a constrained decoder architecture with custom PPO/GRPO losses for better generative performance in GPT-2 models.
- Co-developed and contributed to Lambeq, an open-source modular and composable NLP toolkit with 500+ stars and 80K+ downloads. [1]
Team Lead and Founder
NanoML
Purdue University
2021 - 2024
- Founded a team of 10 researchers developing ML models and algorithms for photonic device design and optimization. Published papers in Applied Physics Reviews, Nature Partner Journal, CLEO, and APS with work funded by NSF, DOE, ORNL (OLCF), and AWS Braket.
- Developed new autoencoder, GAN, and diffusion models for designing meta-optical devices, achieving a 16200% speed-up resulting in a major milestone for the Quantum Science Center with two news articles and 50 citations. [1] · [2]
- Raised $27k in computing resources through grants, saved $40k in ML costs through server hosting, and assisted in raising $200k+ for the QSC Summer School. [1]
- Led the development of attention-based models for tackling the $75B counterfeit chip market leading to a 55,000x speed-up and 41% accuracy increase in the verification of semiconductor devices. [1]
Algorithms Researcher
QuEra Computing
Quantum Machine Learning
2022
- Developed Julia and PyTorch packages for simulating and analyzing the spectral gap of Rydberg atom systems . Demonstrated comparative advantage on neutral atoms for generative AI sampling of photonic devices. [1]
- Set team research objectives and wrote quantum simulation software in Python and Julia to simulate Rydberg atom systems for analysis. [1]
- Verified early-access, open-source, flagship neutral atom simulation software Bloqade on machine learning sampling tasks. [1]
Software Engineer
System-on-Chip Verification
ARM
2019
- Engineered verification software in Python, XML, and C++ for ARMv8 instruction set coverage, leading to a 250% improvement in Chi-Square randomness and immediate positive feedback from customers for improving test consistency.
- Rebuilt 2 ARMv8 debug instruction tests in ARM's verification suite to reduce the complexity of debugging for customers.
- Designed a 32-Bit MIPS RISC processor in Verilog with L1-Cache and 5-stage pipeline. Compiled C to MIPS IS and benchmarked on FPGAs.
- Developed Python to RISCV transpiler for generating Verilog Read-Only-Memory for RISCV-based System-on-Chip. [1]
Research interests
- Machine Learning and Generative Modeling for Multiphysics and Design
- Discrete Mathematics & Statistical Mechanics
- Quantum Computing
- Software Engineering & Dev-Ops
Publications
- 1.
Bounds on Decorated Sweep Covers in Tree Posets
Wilson, B.A., Krawchuk, C.
arXiv preprint arXiv:2604.03165
- 2.
Deep learning in photonic device development: nuances and opportunities
Iyer, V., Wilson, B.A., Chen, Y., Kildishev, A.V., Shalaev, V.M., Boltasseva, A.
npj Nanophotonics 3, 5 (2026)
- 3.
PearSAN: A Machine Learning Method for Inverse Design Using Pearson Correlated Surrogate Annealing
Bezick, M., Wilson, B.A., Iyer, V., Chen, Y., Shalaev, V.M., Kais, S., Kildishev, A.V., Lackey, B., Boltasseva, A.
Advanced Optical Materials (2026): e00249
- 4.
Chen, Y., Montes McNeil, A., Park, T., Wilson, B., Iyer, V., Bezick, M., Choi, J., Ojha, R., Mahendran, P., Singh, D.K., Chitturi, G., Chen, P., Do, T., Kildishev, A., Shalaev, V., Moebius, M., Cai, W., Liu, Y., Boltasseva, A.
Nanophotonics, Vol. 14, Issue 23 (July 2025)
- 5.
Authentication Through Residual, Attention-based Processing of Tampered Optical Responses
Wilson, B., Chen, Y., Singh, D.K., Ojha, R., Pottle, J., Bezick, M., Boltasseva, A., Shalaev, V., Kildishev, A.
Advanced Photonics, Vol. 6, Issue 5, 056002 (July 2024).
- 6.
Non-native Quantum Generative Optimization with Adversarial Autoencoders
Wilson, B., Wurtz, J., Mkhitaryan, V., Bezick, M., Wang, S.T., Kais, S., Shalaev, V., Boltasseva, A.
arXiv:2407.13830
- 7.
A Relative Church-Turing-Deutsch Thesis from Special Relativity and Undecidability
Wilson, B., Dickey, E., Iyer, V., Kais, S.
In Peer-Review, arXiv:2206.06419
- 8.
Planning for Package Deliveries in Risky Environments Over Multiple Epochs
Wilson, B., Hudack, J., Sundaram, S.
American Controls Conference 2022, arXiv:2110.09917
- 9.
Wilson, B., Kudyshev, Z., Kildishev, A., Shalaev, V., Kais, S., Boltasseva, A.
Applied Physics Reviews, 8, 041418, (Impact Factor: 19.16)
- 10.
Conference talks & posters
- 1.
Physics-Conditioned Diffusion Model for the Inverse-Design of High-efficiency Thermophotovoltaic Metasurface
Chen, Y., Bezick, M., Wilson, B., Kildishev, A., Shalaev, V., Boltasseva, A.
CLEO (2025)
- 2.
Photonic Inverse Design Through Machine Learning and Correlated Surrogate Annealing
Bezick, M., Wilson, B., Iyer, V., Chen, Y., Shalaev, V., Kais, S., Kildishev, A., Boltasseva, A., Lackey, B.
CLEO (2025)
- 3.
Bezick, M., Wilson, B., Iyer, V., Chen, Y., Shalaev, V., Kais, S., Kildishev, A., Boltasseva, A., Lackey, B.
APS Global Physics Summit March Meeting (2024)
- 4.
Physics-conditioned deep generative model for high-efficiency metasurface thermal emitter design
Chen, Y., Bezick, M., Wilson, B., Kildishev, A., Boltasseva, A., Shalaev, V.
APS Global Physics Summit March Meeting (2024)
- 5.
Machine-learning-assisted optical authentication of chip tampering
Wilson, B., Chen, Y., Singh, D.K., Ojha, R., Pottle, J., Bezick, M., Boltasseva, A., Shalaev, V., Kildishev, A.
Photonic Computing: From Materials and Devices to Systems and Applications (2024)
- 6.
Tandon, S., Welling, A., Wilson, B., Mkhitaryan, V., Shalaev, V., Boltasseva, A.
Purdue University Spring Undergraduate Research Conference (2024)
- 7.
Discrete, Tunable Wavefront Engineering using Binary Quadratic Optimization
Tandon, S., Welling, A., Wilson, B., Mkhitaryan, V., Shalaev, V., Boltasseva, A.
Purdue University Spring Undergraduate Research Conference (2024)
- 8.
Machine learning-assisted rapid clustering and mechanism exploration of single photon emitters
Chitturi, G., Do, T., Chen, Y., Wilson, B., Shalaev, V., Boltasseva, A., Senichev, A.
Purdue University Spring Undergraduate Research Conference (2024)
- 9.
Malavathu, R., Wilson, B., Boltasseva, A.
Purdue University Spring Undergraduate Research Conference (2024)
- 10.
Applications of Variational Neural Annealing for Machine Learning-Assisted Topological Optimization
Bezick, M., Iyer, V., Wilson, B., Boltasseva, A.
Purdue University Spring Undergraduate Research Conference (2024)
- 11.
Machine Learning Realization of PUFS with Random Plasmonic Systems
Singh, D.K., Ojha, R., Chen, Y., Wilson, B., Bezick, M., Boltasseva, A., Shalaev, V., Kildishev, A.
CI + AI Cyberinfrastructure Symposium (2023)
- 12.
Latent Diffusion for Material Topology Sampling
Bezick, M., Wilson, B., Boltasseva, A.
CI + AI Cyberinfrastructure Symposium (2023)
- 13.
Learning Van der Waals Potentials in Surrogate Rydberg Hamiltonians
Wilson, B., Iyer, V., Shalaev, V., Kildishev, A., Kais, S., Boltasseva, A.
3rd Annual Quantum Summer School (2023)
- 14.
Denoising Diffusion for Material Topology Sampling
Bezick, M., Wilson, B., Boltasseva, A.
Spring Undergraduate Research citation (2023)
- 15.
Plasmonic nanoparticle densities for physical verification of unclonable spectral tags in microelectronics packaging
Singh, D.K., Chen, Y., Wilson, B., Boltasseva, A., Shalaev, V., Kildishev, A.
Spring Undergraduate Research citation (2023)
- 16.
Empowering Quantum 2.0 Devices and Approaches with Machine Learning (QTu2A.13)
Wilson, B., Chen, Y., Shalaev, V., Kildishev, A., Kais, S., Boltasseva, A.
Quantum 2.0 (2022)
- 17.
Metasurface Compression Analysis via bVAE Reconstruction Loss
Wilson, B., Iyer, V., Shalaev, V., Kildishev, A., Kais, S., Boltasseva, A.
ECE Elmore Emerging Frontiers Center Poster Session (2022)
- 18.
Machine learning for photonics Active Photonic Platforms
Boltasseva, A., Shalaev, V., Wilson, B.
2022, PC121960T
- 19.
Source Shaping for Electromagnetic Optimization via Higher-Order Variational Quantum Algorithms
Wilson, B., Mkhitaryan, V., Shalaev, V., Kildishev, A., Kais, S., Boltasseva, A.
2nd Annual Quantum Summer School (2022)
- 20.
Machine Learning for Nanophotonic Design and Quantum Measurements
Wilson, B., Chen, Y., Shalaev, V., Kildishev, A., Kais, S., Boltasseva, A.
Purdue Elmore Center (2021)
- 21.
Metasurface Design Optimization via D-Wave based Sampling
Wilson, B., Kudyshev, Z., Kildishev, A., Shalaev, V., Kais, S., Boltasseva, A.
CLEO 2021
Grants
-
AWS Braket Research Grant
Amazon
2023 - 2024
-
ACC 2022 Student Travel Grant
ACC 2022
2022
-
Universities Space Research Association Quantum (Cycle 4)
NASA Ames Center
2020 – 2021
-
Leadership Computing Facility D-Wave
Oak Ridge
2020
Fellowships
-
Summer Undergraduate Research Fellowship
Purdue University
2017
Sponsors & programs
-
Quantum Science Center
Oak Ridge
2021 - 2024
-
Purdue Elmore Emerging Frontiers Center
Elmore Family
2022 - 2024
-
QuEra Computing
Harvard/MIT
2022
-
Air Force Research Lab
AFRL
2021
Leadership
-
Zephram Co-Founder, Former CTO & Interim CEO
Zephram
2025 -
-
NanoML Team Lead
NanoML Team
2021 - 2024
-
Quantum Science Center Summer School Committee
Quantum Science Center
2021 - 2023
-
DOE Quantum Science Center PGA Team Lead
Quantum Science Center
2021 - 2022
-
SURF Graduate Mentor
Purdue SURF
2022 - 2023
-
Elmore Center Poster Session Committee
Elmore Center
2022
-
Graduate Teaching Assistant
Purdue ECE
2020 - 2021
-
SURF Graduate Assistant
Purdue SURF
2020
-
SURF Symposium Planning Committee
Purdue SURF
2020
-
Stories of Success: Fireside Chat Host
Purdue ECE
2019 - 2020
-
ECE Shop Graduate Assistant
Purdue ECE
2019 - 2020
-
Software Team Lead
SoCET Team, Purdue
2018 - 2019
-
Resident Assistant
Purdue University
2020 - 2021
-
Volunteer Robotics Teacher
Muslim Learning Society
2014
-
First Robotics Software Team Lead
Plainfield Earthquakers
2013 - 2015
-
Video Game Lead Developer
Wired Vision Games
2012 - 2015
Speaking & panels
-
Academic Research Panel Member
Qiskit Fall Fest
2023
-
Panel Moderator
QSC Summer School
2022 - 2024
-
American Controls Conference
ACC
2022
-
SURF Graduate Mentor
Purdue University
2020 - 2023
-
Fireside Chat Interviewer
Quantum Science Center
2021 - 2022
-
Stories of Success Fireside Chat Host
Purdue University
2019 - 2020
Reviewing & committees
-
Reviewer
Advanced Photonics
2024
-
Reviewer
Quantum
2022 - 2024
-
Reviewer
Nature Communications
2022
-
Reviewer
Advanced Materials
2022
-
Poster Session Committee
Elmore Center
2022
-
Online Research Symposium
Purdue SURF
2020
Teaching & formal roles
-
First Time Researcher Fellowship Mentor
Purdue University
2022 - 2024
-
Graduate Mentor
Purdue SURF
2020 - 2023
-
Graduate Teaching Assistant (Data Structures and Algorithms)
Purdue University
2021
-
Graduate Assistant
Purdue SURF
2020
-
Graduate Teaching Assistant (Advanced C)
Purdue University
2020
-
Resident Assistant
Purdue University
2016 - 2018
Competitions
-
2nd Place National Rube Goldberg Competition
Purdue University
2016
-
2nd Place Purdue First Robotics Regionals
Purdue University
2014
-
3rd Place IUPUI Mathematics Competition
Indiana University Purdue University of Indianapolis
2014
Additional appointments
-
Graduate Research Assistant (Alexandra Boltasseva)
Purdue University
2019 - 2024
-
Committee Member
QSC Summer School
2021 - 2023
-
Advisor
Qiskit Fall Fest
2023
-
Quantum Algorithms Researcher
QuEra Computing
2022
-
Postdoctoral and Graduate Student Association Team Lead
Quantum Science Center
2021 - 2022
-
Graduate Research Assistant (Shreyas Sundaram)
Purdue University
2021
-
ECE Shop Graduate Assistant
Purdue University
2019 - 2020
-
Verification Team Engineer
ARM
Summer 2019
-
Software Team Lead
Purdue SoCET Team
2018 - 2019
-
SURF Fellow
Purdue University
2017
-
Undergraduate Research Assistant
Purdue University
2016 - 2019
Student highlights & mentorship
-
First Time Researcher Fellowship
Advay Welling
EURO 2024
-
First Time Researcher Fellowship
Sarthak Tandon
EURO 2024
-
First Time Researcher Fellowship
Geetika Chitturi
EURO 2024
-
Qiskit Fall Fest Lead
Vaishnavi Iyer
2023
-
Best Poster Presentation
Daksh Kumar Singh
SURF 2023
-
Summer Undergraduate Research Fellowship
Daksh Kumar Singh
SURF 2023
-
First Time Researcher Fellowship
Daksh Kumar Singh
EURO 2023
-
First Time Researcher Fellowship
Michael Bezick
EURO 2023
-
IBM Watson Quantum Summer Internship
Vaishnavi Iyer
2023
-
Summer Undergraduate Research Fellowship
Vaishnavi Iyer
2022
-
Best First Time Researcher Award
David Czerwonki
SURF 2020
-
Best First Time Researcher Award
Rachel Zhang
SURF 2020
Mentees
-
Advay Welling
Purdue SoCET and AMD
-
Daksh Kumar Singh
Purdue MS student under Boltasseva/Shalaev group
-
Daria Shkel
PhD student @ Cornell
-
David Czerwonki
PhD student and Senior Graduate Professional @ Purdue and EURO
-
Geetika Chitturi
Purdue NanoML and NanoX
-
Lee Dongeun
Plasma Engineer @ VM
-
Michael Bezick
PhD student @ University of Maryland
-
Rachel Zhang
Medical Student @ University of Michigan
-
Rohan Malavathu
Amazon
-
Rohan Ojha
Founding Engineer @ Conway
-
Sarthak Tandon
Purdue and AMD
-
Seoyoung Cho
Korea Advanced Institute of Science and Technology
-
Trang Do
Purdue and ORNL
-
Vaishnavi Iyer
Purdue PhD student at Boltasseva/Shalaev group
Software & engineering projects
-
Web Development
NanoML Team
2023 -
-
Custom Arch Linux Setup
Linux
2022 -
-
Neovim Personal Configuration
Linux
2022 -
-
ClearML Server Management
Linux
2022 -
-
Bloqade Rydberg Atom Simulations
QuEra Computing
2022 -
-
NanoML Experiment Manager
Python
2023
-
OpenAI Automatic Documentation Builder
OpenAI
2023
-
OpenAI + Alexa API Integration
OpenAI
2022
-
CEM Codes (Diffraction, Capacitor)
Purdue ECE
2021
-
ECE Shop Sign (RGB LED + Web Server)
Purdue ECE
2019
-
Bluetooth Controlled Powerstrip
Purdue ECE Embedded Sys.
2018
-
32-bit MIPS Processor from scratch
Purdue ECE 437
2018
-
VLSI Clock-Tree Synthesis Project
Purdue ECE 595
2018
-
Ethernet Packet Sniffer
Purdue ECE 337
2018
-
Multiple Pursuers and Sweep-Covers Enumeration
Purdue ECE Research
2017
-
8-bit Breadboard PC - Synchronized Clock
Personal Hobby
2017
-
Tanvas Image Processing and Material Analysis
Northwestern Hackathon
2016
-
Zombie Video Game
Wired Vision Games
2014
-
Falling Balls Android Game
Personal Hobby
2013
Media & highlights
- RAPTOR takes a bite out of global counterfeit semiconductor market
Purdue Research Foundation
2024
-
2019
Affiliations
-
Optica
Optica
2021 -
-
Quantum Science Center
Oak Ridge National Lab
2020 - 2024
-
Elmore Emerging Frontiers Center
Purdue University
2020 - 2024
-
Purdue Quantum Science and Engineering Institute
Purdue University
2019 - 2024
-
QuEra Computing
Harvard/MIT
2022
-
IEEE Computer Society
IEEE
2020 - 2021
-
Purdue Society of Professional Engineers
Purdue University
2015 – 2016
-
Wired Vision Games
Game Development Group
2011 – 2015
-
First Robotics
NASA
2013 – 2015
-
Freedom Chairs (Charity Wheelchair Repair)
Indiana Non-Profit
2013 – 2015
-
Project Lead the Way
Indiana Non-Profit
2011 – 2013
References
-
Sabre Kais (Purdue)
kais@purdue.edu
-
Alexandra Boltasseva (Purdue)
aeb@purdue.edu
-
Travis Humble (ORNL)
humblets@ornl.gov
-
Sheng-Tao Wang (QuEra)
swang@quera.com
-
Stephen Clark (Quantinuum/Deepmind)
cl.cam.ac.uk
-
Vladimir Shalaev (Purdue)
shalaev@purdue.edu
Research statement
My research spans machine learning, physics-informed modeling, and optimization for engineered systems—from nanophotonics and semiconductor verification to quantum circuit synthesis and multi-agent scientific inference.
Methodologically I focus on scalable algorithms (including CUDA-accelerated workflows), generative and simulation-based modeling, and collaboration across disciplines.
More to come here…
Teaching statement
Statement of Teaching Philosophy
Blake A. Wilson
The value of an engineer is fundamentally shifting. In an era where generative AI and automated solvers can rapidly produce code, circuit schematics, and system architectures, the primary role of the engineer is not primarily a designer, but a verifier. My overarching goal as an educator is to train students to be “critical designers” who can rigorously validate automated outputs and understand the physical and computational layers from first principles. To achieve this, my pedagogy relies on testing foundational knowledge in exams, active failure analysis when interacting with generative tools, and inclusive mentorship that treats students as junior colleagues.
The “Verify-First” Classroom
When students step into the modern engineering landscape, they are handed tools that act as faulty design generators. If we don’t teach students how to criticize generative tools and they rely on them too much to retrieve information, they will be ill-equipped to catch the subtle, systemic errors these tools introduce. Therefore, I utilize a backward-design process to build courses where systems orchestration and failure analysis are primary learning outcomes.
In practice, this means moving beyond traditional rote problem sets. I implement active learning frameworks such as adversarial justification. For example, in an algorithms or architecture course, I might provide students with a faulty AI-generated script designed to optimize a data structure or simulate a hardware component. The assignment is not to write the code from scratch, but to act as the senior engineer: they must write the test bench that breaks the AI’s code, mathematically justify why the AI’s solution is suboptimal, and provide a correct modification. This forces students to deeply engage with the underlying logic rather than passively trusting a “black box” output.
Scaffolding Mental Models: Tools vs. Truth
To verify generative tools, students must first possess a rock-solid internal mental model of the fundamentals. I achieve this by distinctly separating how tools are used in preparation versus how mastery is validated:
- Homework as Orchestration: I design homework assignments as a training ground for real-world engineering. Here, students are encouraged to use modern tools to build intuition, explore edge cases, and piece together complex workflows. This teaches them how to use the tools effectively and safely.
- Exams as Foundational Validation: Conversely, I maintain a strict “no-tool” environment for foundational exams. This is not to artificially restrict students, but to accurately assess their individual mastery of core concepts—such as memory management in C or the physics of an electrical circuit. If a student does not possess the foundational truth in their own mind, they cannot act as the “human-in-the-loop” when a tool inevitably hallucinates.
Mentorship and Inclusive Engineering
Beyond the lecture hall, my teaching philosophy is heavily informed by my experience leading cross-functional research teams across Oxford, Cambridge, and Harvard. I have formally mentored over 200 researchers, and my approach is always to demystify the “hidden curriculum” of STEM. I focus heavily on creating a welcoming environment for students from non-traditional backgrounds, helping them connect with the engineering community and its principles no matter their background.
I do this by treating my mentees as junior colleagues from day one. Instead of handing them solved problems, we collaborate on the frontier of what is unknown. We discuss not only the technical mechanisms of our work but the broader societal and ethical landscape of computing. By empowering students to bridge the gap between disciplines and giving them ownership over their work, they transform from students into confident, capable engineers.
Ultimately, my goal is not just to teach students how to build systems, but to ensure they have the technical rigor and ethical ownership to take professional responsibility for the next generation of engineering.