Research Areas

The Hybrid Optoelectronic Correlator (HOC) was proposed and demonstrated at LAPT. I was the first to show it is capable of producing shift, scale, and rotation invariant (SSRI) target recognition. The maximum theoretical operating speed is on the order of 5 µs, making it a viable input filter for more resource-intensive image processing systems.

The HOC uses optics to perform real-time 2D Fourier transforms of arbitrary images. Magnitudes are captured directly with focal plane arrays; phases are captured using off-axis auxiliary plane waves. These signals are processed electronically (added, subtracted, and multiplied pixel-by-pixel) and the result is sent to a DMD SLM, which produces the 2D cross-correlation via a second Fourier transform. This architecture is inherently shift invariant. Scale and rotation invariance are achieved via the Polar Mellin Transform (PMT), computed in C++ in <900 µs for Full HD images, the fastest on record.

I proposed the Balanced Joint Transform Correlator (BJTC), which merges techniques from the HOC and the traditional JTC. This design achieves SSRI recognition with half the optical complexity of the HOC and orders-of-magnitude higher SNR than a standard JTC. The BJTC uses phase information (HOC) to eliminate the self-correlation bias terms: BJPS = JPS − |R|² − |Q|². Simulated improvement: ~800×; experimental: 8.4× peak enhancement. Operating at 720 fps, it is approximately an order of magnitude faster than current state-of-the-art computational systems.

A novel correlator architecture that applies the Mellin transform along the temporal axis (rather than the spatial axes), enabling speed-invariant event recognition from 1D time-domain signals. The approach uses a spectrogram (STFT) to lift time-domain signals into 2D space, then applies the HOC pipeline, producing recognition invariant to playback speed. Demonstrated on audio signals (speaker-invariant "Northwestern" vs. "Northwestern University") and AER-encoded video. An atomic medium photon-echo implementation supports ~18 unique nonlinear signal components for enhanced discrimination.

The HOC has evolved from early prototypes using a Thorlabs DCC1545M camera (1280×1024, 30 fps) and a TI DLP3000 SLM (684×608, operating at 854×480, ~6.5 mm physical DMD) with a Verdi V2 laser, through to the current implementation using XIMEA LUX19HS cameras (2200 fps, 4 units) and TI DLP471TE SLMs (720 fps, 8-bit), driven by custom C++ with OpenCV on a multi-core CPU. An earlier FPGA prototype used a Xilinx VCU118 with ONSemi focal plane arrays and 4 VHDL modules (UART Rx, RAM, ROM for LPT mapping, UART Tx).

My PQ:PMMA research has focused on fabrication, characterization, and applications of this holographic polymer. Unlike most holographic materials, it does not suffer from post-exposure shrinkage and can be manufactured at centimeter-scale thickness, enabling self-sustaining HOEs with extremely high angular and spectral selectivity (Bragg selectivity). I've achieved Δn slightly above 10⁻⁴, among the highest reported for this material. My two-level model of electric susceptibility (R² = 0.9956 at 780 nm; 0.9903 at 1300 nm; 0.9876 at 1550 nm) enables accurate Δn(λ) prediction from a single UV-Vis measurement.

Demonstrated WDM/DeMUX systems at 780 nm (3-channel: 760/780/795 nm) and 1550 nm (3-channel and 6-channel: 1548.1–1560.27 nm, 2.8 mm aperture, 6 multiplexed messages). The 6-channel 1550 nm system transmitted six independent messages in parallel, all received without error. Theoretical free-space range: 525 km at 1 W / 117 km at 50 mW.

A 2 mm thick PQ:PMMA disk can store 1,320 images (12 angular positions × 110 multiplexed images) using the automated writing arm (DMD-based, two rotation modes). Disk capacity analysis: 250,000 images / 576 GB in a 10 cm disk; angular selectivity 14.7 mDeg for 2 mm substrate. Stored targets spanned several classes of aircraft, spacecraft, and maritime vessels, along with standard resolution charts. Readout power: 2.8 W at 5 µs / 0.467 W at 30 µs.

Designed a compact passive ranging system using a 3-band HOE as a wavelength filter for the oxygen A-band (762/750/780 nm). The HOE simultaneously measures three spectral channels, so T_turb cancels in the intensity ratio I(762)/I(750), an advantage impossible with sequential filter switching. Demonstrated at 100 m and 200 m in a hallway (51.6/43.3/41.8% DE).

This work was carried out in collaboration with the Center for Nanoscale Materials (CNM) at Argonne National Laboratory. The long-term goal is an on-chip fast-light interferometric optical gyroscope and accelerometer operating near rubidium transition wavelengths, building on free-space fast- and slow-light sensors previously demonstrated at LAPT. AlGaAs was chosen for its tunability across this range. Four Argonne CNM facility proposals were accepted to support this work.

My fabrication runs produced the first on-chip lasing devices demonstrated in the group, using distributed Bragg reflector laser designs operating in the near infrared. Other demonstrated components include Y-junctions, passive waveguides, ring resonators, Fabry-Pérot lasers, and integrated detectors.

I developed the group's end-to-end fabrication process from scratch, spanning e-beam resist preparation, e-beam lithography at the CNM, hard mask deposition, dry etching, p- and n-metal deposition with thermal annealing, and final lapping, cleaving, mounting, and wire-bonding.

I also explored acousto-optic ring isolators on thin-film lithium niobate on insulator, an approach that avoids the need for magnetic garnet isolators, alongside an initial look at magneto-optic isolator materials. This strand of the work centred on coupled-wave analysis for ring isolator design and grating coupler efficiency.

I developed a selective QWI process based on a masked rapid thermal anneal, achieving targeted disordering without significantly affecting neighbouring active regions. QWI allows active (lasing) and passive (low-absorption) waveguide regions to be combined on a single chip, which is essential for the target gyroscope architecture.

At Swave Photonics (Apr 2026 to present) I am developing a reference design for a holographic optical projector for display glasses. The work combines computer generated holography with photopolymer HOE fabrication and experimental automation, and I presented prototype work at SID Display Week 2026.

Previously, at Meta's Reality Labs Research (OPALS Team), I focus on the design, simulation, and fabrication of holographic optical elements (HOEs) with complex optical functions for next-generation augmented reality displays. My work builds on the volume holography expertise developed during my PhD, extending it to wearable form factors and production-oriented manufacturing requirements.

I constructed a unique, fully automated holographic exposure and metrology system combining optical, mechanical, and software control, enabling repeatable, high-precision HOE fabrication at scale. I developed Zemax OpticStudio, MatLab, and SolidWorks models of complex diffractive systems, and produced white papers driving the down-selection of three product architectures.

Improved key performance indicators by up to 10× through wave-optical design optimization. Produced proof-of-concept wearable prototypes during my internship (Jul–Oct 2024) that were tech-transferred into a direct product candidate, the only such transfer from this department to date.

Zemax OpticStudio (wave-optic and ray-optic propagation), SolidWorks (mechanical design and tolerance analysis), MatLab (automated data acquisition, analysis, and simulation), coupled-wave theory (Kogelnik's formalism and full-wave extensions), and direct holographic exposure with controlled environmental conditions.

Note: Only publicly available information is described here. Confidential technical details regarding Swave Photonics and Meta products are not disclosed.