A
Asti Dhiya Anzaria
Asti (she/her) · AI/ML Engineer, Android Engineer & Researcher · Indonesia
About

Software engineer and researcher building applied AI/ML systems, RAG pipelines, NLP recommendation engines, and agentic LLM workflows. 3+ years experience in Android engineering at Tokopedia. MSc in Advanced Computer Science (Merit) from the University of Sheffield on a full LPDP government scholarship. Research interests sit at the intersection of AI, digital governance and public infrastructure.

Projects
2026
Asian drama persona based on photos done try it ↗ Drop your selfie into a CLIP model and find the drama universe you were always meant to be in.
What is your K/C-drama persona based on your Instagram?*
CLIP (ViT-B/32) LangChain ChromaDB Gradio Gemini Python
2026
Mood-based food recs in London done try it ↗ Tell the app you're "feeling blue," it finds comfort food. Built with RAG, Pinecone & Gemini to match emotional queries to restaurant atmospheres. No more scrolling, just eating. Modern upgrade to my 2019 mood-based recommender.
Tell me how you are feeling right now.
Sentiment Analysis Pinecone LangChain BERT Python
2026
TaySwift audio aura done try it ↗ Ever wondered what colour your favourite Taylor song would actually be? I turned Spotify audio features into 4 distinct “Audio Auras” to map the sonic DNA of her entire discography.
Inspired by Spotify Wrapped Audio Aura, but make it swifties.
Python scikit-learn K-means
2019
Mood based food recs in Malang done find the journal here ↗ Built an Android app that reads your face and tells you what to eat. Sad? Comfort food. Happy? Healthy vibes. Azure Face API + Certainty Factor = 79.4% accuracy, 83 SUS score. No more "idk what to eat" crises.
Azure Google Maps Certainty Factor Android Java · Python
2026
What drama that keeps you awake during Covid-19? in progress Analysed user reviews + drama metadata to uncover emotional trends in K-dramas during 2020-2022.
Sentiment Analysis NLP BERT Python
2026
What Spotify song are you based on your heartbeat? upcoming Heartbeat, classify emotion, find a matching song in Spotify. A real-time system that recommends songs based on your physiological state. Because sometimes you need BPM that matches your BPM.
Spotify Camera (PPG) Python
Work Experience
2025 — Now
Research Fellow & Co-lead at Mata Garuda Institute Digital Governance, Technology & Innovation Cluster · LPDP Alumni
Co-leads a research cluster identifying digital governance and AI policy issues, facilitating knowledge-sharing sessions across the LPDP alumni network.
2021 — 2024
Software Engineer, Android at Tokopedia Jakarta, Indonesia
Delivered features across Travel, Digital Products, Live-Streaming, and Stories for tens of millions of users. Eliminated critical memory leaks, built multi-layer test architecture, automated OKR reporting with Python + AppScript.
2020
Android Engineer Intern (DANApprentech) at DANA Indonesia Jakarta, Indonesia
Built production-grade Android features in a regulated fintech environment using MVP-Clean Architecture (Dagger2, Retrofit, RxJava).
Additional Experience
2022 — 2024
Instructor & Mentor, Bangkit Academy by Google, GoTo, Traveloka · Indonesia
Instructed 50+ students in Android development and AI/ML. Supervised final project teams from prototype to demo day.
2024
Android Mentor, GenerationGirl Summer Club Indonesia
Designed and delivered a one-week Android front-end workshop for secondary school students with 1:1 mentorship.
2023
Speaker (Android), GoTo DevCamp Android testing workshop for engineering cohorts
2022
Instructor, Tokopedia Academy Multi-layer Android testing strategy (unit + instrumentation)
Education
2024 — 2025
MSc Advanced Computer Science (Merit) University of Sheffield · Full LPDP Government Scholarship
Dissertation: Usability and Environmental Evaluation of User Authentication for Digital Government Services.
2016 — 2020
S.Kom — Informatics Engineering (GPA 3.66/4.00) Universitas Brawijaya · Specialisation: Mobile, Multimedia & Games
Thesis: Mood-based culinary recommendation app — end-to-end Android development from concept to deployment.
Contact