Chris Lomont, PhD

Chris Lomont, PhD

Research Engineer & Inventor · AI/ML, Graphics, Algorithms, Robotics

Summary

Research engineer, mathematician, and inventor who takes hard problems from first derivation to shipped product. At Meta I build foundational graphics technology for Instagram and the Edits video editor: a graphics-effects language, vector rendering and animation engines, HDR color tooling, and image-quality detection. Before that, principal investigator winning and leading funded research in machine learning, 3D reconstruction, robotics, and computer security, and twenty years building a consumer hardware company I co-founded. PhD in mathematics; degrees in computer science, physics, and mathematics; several patents; widely used open-source algorithms.

Work Experience

Software EngineerMeta (Instagram and Edits)RemoteFeb 2023 – Present

Graphics, imaging, and animation engineer on Instagram (2B+ daily users) and the Edits video editor (50M+ users), building foundational graphics layers used across Meta's family of apps.

  • Designed and built a cross-platform graphics-effects language and C++ runtime for mobile video creation: script-driven render graphs, parameter reflection, live editing, and native GPU integration, so new effects can be authored without a native implementation.
  • Designed and built a portable C++ vector rasterizer and a specialized text and vector animation engine for mobile video editing.
  • Designed and built the color lookup-table tooling used to bring HDR and wide-gamut color filters to Edits, and the shared animation mathematics library behind Edits keyframes.
  • Developed low-level Android graphics infrastructure for Instagram's text animations in Reels and Stories, resolving the threading, timing, and memory problems that blocked launch.
  • Invented low-overhead statistical image-quality detectors for black frames and blur in mobile video-transcoding pipelines, using GPU sampling and CPU scoring.
  • Implemented and optimized GPU super-resolution for mobile media creation, using quantization, custom compression, and robust model encoding.
  • Created and run a biweekly engineering seminar on algorithms, graphics, AI, and systems; mentor engineers and give regular talks on color science, HDR, and image quality.
Principal InvestigatorCybernet SystemsAnn Arbor, MI (remote)Oct 2019 – Feb 2023
  • Principal investigator on funded research in machine learning, 3D reconstruction, GPU computing, and computer security; wrote about 25 proposals in three years, won DHS and Army awards, and led the technical work.
  • Built the company's machine-learning practice: deep learning for firearm make-and-model detection in video, IMU path reconstruction in GPS-denied areas, gesture recognition for augmented reality, and touchless fingerprint matching against legacy databases on mobile devices.
  • Technical lead for a handheld real-time 3D scene-capture system for military site documentation ($850K program): SLAM and CUDA surfel fusion on an NVIDIA Jetson to 1 cm accuracy, a custom OpenGL renderer, and an RGB-D compression algorithm I invented to fit the wireless link.
  • Solved GPS-denied tracking of a hand-held mine detector with deep-learned inertial odometry, designing both the neural approach and the ESP32 / RTK-GPS / IMU hardware platform used to collect ground truth.
Senior Principal EngineerLogikosFort Wayne, INNov 2017 – Oct 2019
  • Technical lead for a venue acoustics design tool for a major audio company: wrote the custom 3D rendering engine and an audio path-tracing engine that predicts speaker coverage and guides speaker-array layout (C++, Qt, Windows, and macOS).
Co-founderHypnocubeAnn Arbor, MIJun 2005 – Present
  • Co-founded and run a company (hypnocube.com) that designs, manufactures, and sells LED art and lighting products (Hypnocube, HypnoSquare, HypnoLight, HypnoLSD); over $2M in sales through retail and direct channels.
Principal InvestigatorCybernet SystemsAnn Arbor, MIMay 2003 – Apr 2013
  • Authored, won, and led Phase I and Phase II SBIR research programs worth about $4M, as principal investigator.
  • Security research that produced patents: hardware-assisted rootkit detection by monitoring OS structures over PCIe (FPGA), a USB firewall, automated software vulnerability detection, IPv6 port-hopping protocols, and malware tracking through virtual-machine deltas.
  • Algorithms research: quantum image processing, wavelet compression of NASA satellite imagery, a real-time CSG ray tracer for missile-defense simulation, robot localization and mapping (SLAM), and ad-hoc networking protocols.
Consultant and Game DeveloperVarious clients
  • Consulting for the financial, oil and gas, video game, and visual-effects industries: mortgage-pool pricing optimization, portfolio and oil-well data systems, 3D game technology, and a cloud image-compositing renderer.
  • Game development for several studios: lead programmer of a console game based on a major film (75,000 lines of assembly), subdivision-surface and NURBS technology for real-time 3D titles, network play, and console modeling tools.

Skills

Languages: C++, C, Python, C#, F#, Julia, Kotlin, JavaScript, Go, GLSL, CUDA, Mathematica, and many others

Assembly and low level: assembly for x86/x64, ARM, MIPS/PIC, Xtensa, and many others; reverse engineering and disassembly on many platforms

AI / ML: deep learning, computer vision, super-resolution, model compression, scientific machine learning, training-data generation, identifying and investigating novel uses of AI

Graphics and imaging: GPU rendering and shaders, color science and HDR, image and video pipelines, ray tracing, 3D reconstruction and SLAM, scientific computing and visualization

Algorithms and mathematics: algorithm design, numerical methods, compression, signal processing, cryptography, language and compiler design, quantum computing, competitive programming

Systems: Android, iOS, Windows, Linux; embedded (ESP32, PIC, FreeRTOS, NVIDIA Jetson); FPGA; high-performance code on many platforms (SIMD, GPU, multicore, embedded)

Education

Ph.D., Mathematics, Purdue University
  • Dissertation: "Error Correcting Codes on Algebraic Surfaces".
M.S., Mathematics, Indiana University–Purdue University Fort Wayne
B.S. in Physics, Mathematics, and Computer Science (triple major), Oral Roberts University

Patents

  • US 8,806,619: System and methods for detecting software vulnerabilities and malicious code (granted 2014)
  • US 8,646,082: USB firewall apparatus and method (granted 2014)

Selected publications and open source

  • "The Hidden Subgroup Problem: Review and Open Problems", arXiv quant-ph/0411037, 2004 (100+ citations)
  • Beach, Lomont, Cohen, "Quantum Image Processing (QuIP)", IEEE Applied Imagery Pattern Recognition Workshop, 2003 (100+ citations)
  • "Fast Inverse Square Root", 2003: the analysis of the Quake III constant that the Wikipedia article on the method is built around
  • Five chapters in Game Programming Gems 5–8 (2005–2010): geometric algebra for graphics, floating-point tricks, random number generation, subdivision surfaces, secure channel communication
  • "Introduction to x64 Assembly" (2009) and "Introduction to Intel Advanced Vector Extensions" (2011), written for the Intel developer network and widely used as course material
  • Open source: widely reused algorithm code for AES encryption, FFT, random number generation (WELL512), SSIM image comparison, and posit arithmetic (LibPosit), at github.com/ChrisLomont
  • More papers, articles, and software at www.lomont.org