Jae-Woo Choi, Ph.D. Email & Phone Number

Menlo Park, CA, US

• Support machine learning engineers and research scientists in the Monetization Ranking org to increase Meta’s ads revenue with: (1) innovative modeling techniques (transfer learning, knowledge distillation, etc) (2).

Boston, MA, US

• Exploit the latest multi-task learning approaches to solve core product substitutability, complementarity, deduplication, and competitor matching needs.
• Drove increased adoption of content based product similarity embeddings across personalized product sort, day zero, and personalized product recommendations for batch, low latency, and real time.
• Inception of product substitutability and complementary embeddings for geographically optimized product sort and inventory positioning.
• Define next-gen product matching to unify the algos and systems that deduplicates the Wayfair catalog, matches products to competitor catalogs, and identifies high potential selection gaps.

Boston, MA, US

• Iterate beyond computer vision models to incorporate NLP techniques and solve core product catalog tagging needs to drive product similarity and matching models.
• Drive platform team to unify data science solutions from multiple POCs / MVPs across different BUs to a comprehensive data science product that scales models to the entire catalog.

Boston, MA, US

• Broadening scope from computer vision into speech to get the customer what they want from oral communication.
• Continue leading the computer vision team which aims to get the customer what they want without needing words. For a look at what we do in this space, take look here.
• Owned Wayfair's Computer Vision, Speech, and Language Modeling roadmap and led new high impact and high value initiatives while growing the team from 10 to 30+ product and project managers, data scientists, machine.

Boston, MA, US

• I lead the computer vision team which aims to get the customer what they want without needing words. For a brief look at what we do in this space, take look here.
• Owned Wayfair's Computer Vision roadmap and led new high impact and high value initiatives while growing the team from 1 data scientist to 10 data scientists and machine learning engineers.

Manassas, VA, US

• Develop automated flight system by perceiving aircraft instruments and actuating the controls (http://www.aurora.aero/robotic_copilot/, http://www.linkedin.com/hp/update/6194242271819423744)
• Propose, win, and execute an AI-enabled modular intercept drone using active stereo vision (http://aurora.aero/counter-uas-cuas)
• Program manager and principal investigator on sense-and-avoid on next generation aerial vehicles. Presented research findings at ICAS Workshop on Intelligent and Autonomous Technologies in Aeronautics in 2017.
• Generated and presented roadmaps on perception for autonomous flight to Boeing throughout 2017 as part of acquisition. Serve as member on Boeing's A.I. committee shaping autonomy strategy for next gen aircraft.
• Grew the perception team over 2 years from 2 perception engineers to over 15 software and perception engineers, housing 4 interns during the summer, supporting 15 different programs.

Lake Mary, FL, US

• Develop automated 3-D metrology solution based on structured stereo-vision using two cameras and one projector; successfully launched Cobalt (http://cobalt.faro.com) within 2 years.
• Perform Monte Carlo simulations on error budget of camera imaging system and choose optoelectronic component based on system requirements to minimize costs and risks. Improved the accuracy to sub-0.2 pixels and.
• Develop a novel computational photography algorithm add-on to existing camera calibration, implemented faster image filtering and optimized stereo-vision correspondence algorithms in MATLAB (GPU) and C++/OpenCV/openCL
• Lead prototyping effort on firmware side, writing more than 50% of firmware code to interface with cameras and electronics. Interact with vendors, identifying critical bugs, and implementing workarounds.
• Perform optical characterization of different lenses using Optikos LensCheck. Verify multiple camera sensors against home-built EMVA1288 MATLAB test. Define mechanical tolerances on mount lengths and optical axes.
• Use statistical data analysis in JMP from SQL database to determine effective corrective actions on production floor to tighten tolerances and improve the production yield.

• As a co-founder for 2 different startups, I searched for venture capital partners, wrote business proposals, brought in funding, established IP through patents, led academic and industrial research and development.
• On one project, I wrote an academic research proposal with 4 external collaborators and received a 4 year, 6 million USD project to explore converting sunlight into useful chemical energy. Assigned as the lead.
• Patented and developed a method to ablate biological material with a ultrashort, high powered laser delivered using an endoscope.
• Patented and developed a 1cm2 microfluidic device with 6.5 times higher water purification efficiency compared to other techniques using solar energy.

Pasadena, CA, US

• Assigned as the sole investigator for component development, I eliminated the need of external pumps for lab-on-a-chip devices. I developed optical systems, made microfluidic devices, modeled the system in simulations.
• As the senior researcher of the laboratory, I oversaw equipment purchasing, performed chemical safety checks, and maintained the website. I automated procedures of the lab resulting in 50% less administrative work.

Pasadena, CA, US

• As the junior researcher in the lab, I created a device to diagnose urinary tract infections at home. I made the device using cleanroom technologies, modeled the optical scattering in simulations, and wrote code to.

Falls Church, VA, US

• As the sole firmware programmer on a tiger team of 3, I wrote the device communication component of the software for an outer space data storage unit. I developed, wrote, and tested the code in C++. By working.

Pasadena, CA, US

• As the sole programmer working with PhD students, I implemented a model of mobile ad-hoc wireless networks. I developed, wrote, and tested the code in C++, decreasing the algorithm complexity by 3 orders of magnitude.

Pasadena, CA, US

• As the sole programmer working with scientists, I improved data calibration of synthetic aperture radar generated altitude maps. I developed, wrote, and tested the code in IDL, which calibrated peaks and valleys to.

Los Angeles, CA, US

• As the sole intern working with a geologist consultant, I mapped the geologic area of Los Angeles. I compiled and input the data into FORGAS software, resulting in a 3-D model to forecast the economic profits of.

Phd, Electrical Engineering

Ms, Electrical Engineering

Bs, Electrical And Computer Engineering

Education record