LLM systems | Computer vision | Applied machine learning
I ship ML systems end-to-end, from evaluation and grounded generation to production-minded computer vision and data products.
Portfolio | LinkedIn | Hugging Face | Email
| Area | Focus |
|---|---|
| LLM systems | Evaluation, grounded generation, RAG, citations, and provider-agnostic integrations |
| Computer vision | Detection, OCR, document understanding, and image-based automation |
| Applied ML | Geospatial modeling, forecasting, tabular models, and competition-grade validation |
| ML products | APIs, interactive demos, full-stack workflows, and production-minded tooling |
The projects below reflect how I like to work: start with a real constraint, make the model traceable, and ship an interface someone can use.
|
Research papers, turned into cited conversations. Fetches a paper, writes a two-host dialogue, verifies every claim against its source, and produces audio, transcripts, subtitles, show notes, and optional video.
|
Lecture recordings, turned into study material. Creates searchable transcripts, grounded summaries, structured notes, key concepts, and adaptive quizzes with private user workspaces and six swappable STT providers.
|
|
Precipitation estimates from satellite imagery. A satellite-only Solafune competition baseline using Himawari, GOES, and Meteosat imagery, a compact scratch U-Net, geographic validation, calibration, and validated 41 x 41 GeoTIFF outputs.
|
Handwritten election forms, read by computer vision. Detects vote regions with YOLOv8 and recognizes handwritten digits with fine-tuned TrOCR through an interactive Streamlit app. 88.7% mAP50-95 | 0.8% CER | 1st place, Gammafest Data Science Competition 2024 |
| Domain | Tools |
|---|---|
| Languages | Python, R, SQL |
| ML / deep learning | PyTorch, TensorFlow, Keras, scikit-learn, LightGBM |
| LLMs / NLP | Transformers, IndoBERT, Llama, LangChain, RAG, ChromaDB |
| Computer vision | YOLOv8, YOLO11, TrOCR, PaddleOCR, MobileNet, OpenCV |
| Serving / data | FastAPI, Flask, Streamlit, pandas, NumPy, Matplotlib, Tableau |
View my full contribution activity or explore my repositories.
I am open to AI/ML Engineer and Data Scientist opportunities, collaborations, and thoughtful technical conversations.
Email: rachmadiapri@gmail.com
Portfolio: aprirachmadi.github.io