Bryan Dale.

pelitacargo-track-dashboard.vercel.appOpen ↗

A shipment tracking dashboard for China–Indonesia freight forwarding, modeled on how Pelita Cargo works. Customers search the manifest and follow each shipment through six stages, from pickup to delivery; admins add shipments and move them forward. Data lives in Supabase, and the cost estimator uses Pelita Cargo’s real all-in rates — per CBM for sea freight, per kg for air.

• TrackSearch by waybill or client and see a timestamped stage timeline.
• AdminAdd shipments by marking code, volume and rate, then update their status.
• EstimateCalculate all-in shipping cost from real per-CBM and per-kg rates.

Stack

React 19ViteReact RouterSupabasePostgreSQLCSSVercel

gastro-tmb-tmb-prediction.vercel.appOpen ↗

A screening tool that predicts whether a gastric cancer tumor is TMB-High or TMB-Low from the mutation status of 10 genes. A Random Forest trained on 774 patients from five cBioPortal cohorts powers a bilingual (Indonesian / English) web app that shows the prediction’s confidence, each gene’s contribution, and a therapy recommendation.

1. 5 cBioPortal cohorts
2. 10-gene mutation features
3. Random Forest
4. Flask backend
5. TMB-High / Low + therapy

Results

90% test accuracy · F1 94% on TMB-Low, 62% on TMB-High

Stack

PythonFlaskscikit-learnpandasHTMLCSSJavaScript

An AI-powered foot measurement system for shoe sizing. A photo with a reference object is calibrated with homography, then deep-learning segmentation and anatomical landmark detection turn pixels into real-world measurements — served as a FastAPI microservice and deployed with Docker.

1. Photo + reference object
2. Homography calibration
3. Foot segmentation
4. Landmark detection
5. Foot measurements

My part

Collected the photo dataset, trained the model, helped develop the model, and worked on feature extraction.

Stack

PythonPyTorchOpenCVFastAPIDocker

Named-entity recognition for crypto assets in Indonesian tweets. Training labels come from distant supervision, and heuristic-augmented denoising cleans them before training an IndoBERTweet encoder with a CRF layer.

1. Indonesian tweets
2. Distant supervision labels
3. Heuristic denoising
4. IndoBERTweet-CRF
5. Crypto entities

My part

Worked with the team through the pipeline — preparing the tweet data, distant-supervision labeling, heuristic denoising — and co-wrote the paper.

Stack

PythonJupyterIndoBERTweetCRF