Summary
I am a 4-year-experienced NLP-antusiast Software Engineer who is specialized in back-end engineering. I enthusiastically learn, and explore many new things as a part of my curiosity. I enjoy spending my off-work time doing side projects, which helps me to catch up with new technologies, as well as to improve my software engineering excellence.
Education
| Duration | Qualification | Institute/School | Result |
| 4 years | Bachelor of Computer Science | Bandung Institute of Technology |
|
| 3 years | Natural Science | SMAK 1 BPK PENABUR Bandung | - |
Working Experience
Sembly, Singapore
Sembly is a multi-national company which is building a collective braintrust social media. Sembly aims to be the leading platform for providing the most accurate and trustworthy information, which is achieved by facilitating the most well-reviewed collaborative discourse.
- Belong to the search team, to implement all 10+ search-related features in the backend service, which integrates Algolia & ElasticSearch as the core internal search engines.
- Develop both classical & semantic-based search, which utilizes lots of search technologies such as: BM25, optimized roaring bitmap, sentence encoding, semantic vector matching, as well as modern LLMs like GPT4, and Mixtral8x7B.
- Maintain & design performance vs cost trade-off of the search index, picking the right trade-off for various corresponding use cases, which services are deployed as microservice, preparing for usage by 100+ million concurrent users.
- Lead the GlobalSearch & LocalSearch project, which creates the mostly-used search as an entrypoint search in Sembly. The searches integrate most of ElasticSearch features, retaining a low internal latency (<100 ms) for a whole database search.
- Accelerate sentence-encoding and document indexing process by around 50%, and significantly reduce maintenance and future development cost.
- Mainly use Golang, ReactJS, Algolia, and ElasticSearch.
Prosa.ai, Bandung
Prosa.ai is an Indonesian AI-based company, which sells AI-as-a-service. Prosa.ai is a both research & profit company, which always tries to innovate and contribute in latest research, as well as practising the research in client’s real-world use cases.
- Mainly developed colloquial word normalization model, including serving it as an API, developed using library OpenNMT.
- Maintained 10+ existing Prosa.ai's NLP toolkit, including deploying them: from off-premise services by utilizing Gitlab CI/CD, to on-premise services by designing various automated scripts, Daemon service, Docker, and Makefile.
- Mostly used Python, using various deep learning libraries like Tensorflow and PyTorch.
Garasi.id
Garasi.id is a car service providing platform, which also used to sell 2nd or 3rd-hand cars. Garasi.id already has partnerships in more than 30 cities across Indonesia.
- Developed car listing moderation service which utilizes ML-based car image recognition, and was deployed as a performant microservice.
- Experimented with lots of Git features, which involves heavy Git history rewriting.
- Mainly used Python, and AmazonAWS.
Side Projects
Private Music School ERP (Demo link)
- A simple ERP (Enterprise Resource Planner) to manage daily administration tasks of a private music school.
- Technology used: Golang (backend), Next.js (frontend).
Church Liturgy Presentation Automation Framework (Github link)
- An automatic liturgy (PDF) extractor, to be converted into OpenLP-friendly format. The converted result will later be edited in a Word-like editor to adjust the final layout. The result can be used directly in OpenLP stages to display custom layout for weekly Sunday Services.
- Technology used: JavaScript, CSS, Python, HTML5, open source software OpenLP.
Student-to-Supervisor Thesis Assigner (Github link)
- A Google App Script program to help the assignment process of thesis students to their supervisors using Google Spreadsheet and Google Form.
- Technology used: Google Apps Script (Sheet, Form)