LTDK

Résumé

TLDR: shorter ver­sion

Experiences

Now

I’m a Senior Machine Learning Engineer at Visenze, where I fo­cus on mak­ing prod­uct search bet­ter:

• Re-designed Multisearch—crafting new search logic and de­vel­op­ing fea­tures to en­hance end-user ex­pe­ri­ences: rerank & rec­om­men­da­tion ser­vices, nat­ural fil­ter­ing, boost­ing mech­a­nisms, adap­tive thresh­old, and au­to­com­ple­tion. I also lead the ef­fort to im­prove search qual­ity, span­ning from im­ple­ment­ing al­go­rithm code­bases, ex­per­i­ment­ing with and fine­tun­ing text em­bed­dings, to de­sign­ing vir­tual agents for an­no­ta­tion tasks.

• Vector search en­gine: con­ducted ex­per­i­ments on vec­tor search al­go­rithms, bench­marked search qual­ity of SIMD-based vec­tor cal­cu­la­tion^[1], and tuned op­ti­mal HNSW with repli­ca­tion set­tings.

• Object Detection: over­hauled an­no­ta­tion guide­lines, up­graded the train­ing frame­work from Detectron to PyTorch^[2], and de­vel­oped and re­leased mul­ti­ple de­tec­tion mod­els across all Visenze so­lu­tions (search, tag­ging, rec­om­men­da­tion).

I also spent ap­prox­i­mately two years at Visenze de­vel­op­ing on-de­vice ML so­lu­tions:

1. Built a 3D ob­ject la­belling so­lu­tion com­pris­ing: (1) an AR mo­bile app for cap­tur­ing ob­jects and col­lect­ing 3D data, (2) a Unity tool for im­port­ing and re­fin­ing raw data, and (3) a 3D ob­ject de­tec­tion code­base for train­ing, eval­u­a­tion, and vi­su­al­iza­tion.
2. Delivered vir­tual try-on (a sneak peek), ob­ject track­ing, and hand ges­ture recog­ni­tion so­lu­tions on mo­bile de­vices. I de­vel­oped DL mod­els de­signed for low mem­ory us­age and fast in­fer­ence to main­tain a 30 FPS con­straint, and in­te­grated them into ex­ist­ing Unity games via na­tive plu­g­ins. The game has achieved over 10M down­loads on Play Store (Android) and ranks among the top 30 Education apps on iOS Store.

Other pro­jects I was in­volved in:

• Re-designed the se­quen­tial rec­om­men­da­tion al­go­rithm.
• Improved ex­act prod­uct search via new re-rank logic & ad­vanced data aug­men­ta­tions^[3].

Past

Before join­ing Visenze, I spent 2.5 years as a re­search as­sis­tant in Professor Ngai-Man Cheung’s lab at SUTD, fo­cus­ing on:

• Hashing-based im­age re­trieval: We were among the first to jointly learn hash­ing and vec­tor em­bed­ding in an end-to-end man­ner for re­trieval tasks. I im­ple­mented most of the pro­posed meth­ods on Caffe and Torch (not PyTorch)^[4] and con­tributed to 10 pub­li­ca­tions.
• Vision-based lo­cal­iza­tion: We built a 3D model of Singapore from Google Maps to in­fer user lo­ca­tion based on build­ing pho­tos. I was re­spon­si­ble for re-im­ple­ment­ing the in­fer­ence frame­work (originally in MATLAB) for Android de­vices. The most chal­leng­ing task was im­ple­ment­ing fast SIFT fea­tures on Android us­ing shader pro­grams while main­tain­ing ac­cu­racy.

Education

Now

Since 2023, I have been pur­su­ing Master of Computing (Artificial Intelligence Specialisation) (part-time course­work) at NUS. This pro­gram al­lows me to delve into AI top­ics that I don’t en­counter in my daily work. I’ve com­pleted sev­eral pro­jects:

• Menu Recommendation us­ing Deep Learning Critic-Actor Framework [Github, Report]: Thanks to a team mem­ber, we had ac­cess to or­der his­tory of an F&B fran­chise in Singapore. I led im­ple­men­ta­tion and model train­ing. This pro­ject made me re­al­ize that train­ing an RL model is painful.
• Predicting HDB Rental Price in Singapore [Github, Report]: We stud­ied rental price in­creases in Singapore us­ing 2021 – 2023 HDB data and de­vel­oped sev­eral pre­dic­tion mod­els, achiev­ing #1 on the pri­vate leader­board.
• Blood Cell Identification: I pro­posed a U-Net vari­ant that uti­lizes seg­men­ta­tion masks to clas­sify five leuko­cyte types. It achieves 92.59% ac­cu­racy^[5] on the CAM16 dataset.
• LightningCat [Report]: A cloud-based SaaS plat­form pro­vid­ing real-time up­dates on Singapore’s sport­ing fa­cil­i­ties, in­clud­ing op­er­a­tional sta­tus, ca­pac­ity, weather-re­lated clo­sures, and book­ing slots through an in­tu­itive map in­ter­face to op­ti­mize fa­cil­ity us­age and fos­ter com­mu­nity en­gage­ment.
• TemporalLens [Github, Report]: An AI-driven pipeline that tracks emerg­ing top­ics, con­structs event time­lines and knowl­edge graphs, and an­a­lyzes sen­ti­ment shifts.
• PQ-GCN2: Modernized the PQ-GCN code­base to PyTorch Geometric and ex­plored var­i­ous GCN vari­ants for large-scale node clas­si­fi­ca­tion.

Past

I com­pleted my un­der­grad­u­ate de­gree at HCMUS, where I had the op­por­tu­nity to en­gage in sev­eral in­tern­ships:

• I have 1 pa­per at ICIP on com­pact PCA com­po­nent^[6].
• At JAIST, I con­ducted re­search on back­ground sub­trac­tion meth­ods.
• At University of Saskatchewan, I de­rived a new method for white cell count­ing prob­lem.

My Interests

• Information Retrieval & Recommendations.
• Data struc­tures & Algorithms.
• Computer Vision.
• Computer Graphics.
• Game Development.

Skills

Programming Languages

• Professions: Python, Go, C++, Java, C#.
• Hobbies: Julia, D, Crystal, Raku, Clojure.

Tools

• Professions: PyTorch, HuggingFace, Mediapipe, OpenCV, Sklearn, Unity, Docker, Latex, XGBoost.
• Hobbies: Raylib, ImGui.

Misc

• A col­lec­tion of my fa­vorite books.

Build Status