Bryan Wijaya
AI Engineer
11+ years exp
6+ years ML exp
C++
SQL
JavaScript
Python
C#
Java
Visual Basic
PHP
Node JS
Invite to interview
Download CVCV
Overview
Technical skills
Timeline
Roles
Overview
A generalist developer (middle level) focused on small end-to-end projects with a strength in building interactive C# desktop applications and lightweight LLM-integrations. The strongest proven skill is implementing interactive Windows Forms application logic as demonstrated in SpaceShooters/Form1.cs which orchestrates timers, UI rendering and game state, and building simple RAG prototypes in rag_chatbot.py using Chroma and Ollama. There is little public evidence of production backend practices such as database migrations, test coverage, scalable service architecture or robust observability.
Phone
Technical skills
Languages
9
C++
SQL
JavaScript
Python
C#
Java
Visual Basic
PHP
Node JS
AI/ML
24
NumPy
Pandas
Scikit-learn
ChatGPT
TensorFlow
OpenCV
YOLO
Keras
Google Colab
Jupyter Notebook
Fine-tuning
LLM
Gemini
LangChain
LangGraph
XGBoost
Gradio
Unstructured.io
Transfer Learning
Computer Vision
Reinforcement Learning
MLFlow
Ollama
LangSmith
Databases
7
MySQL
Oracle
PostgreSQL
MS SQL
Chroma
Milvus
Qdrant
Mobile
3
Room
Android SDK
MVVM
Other
21
Matplotlib
Git
Seaborn
PyGame
Docker
CNN
Few-Shot Learning
Image Segmentation
Tesseract OCR
Claude Code
Model Context Protocol
AI Agents
RAG
Prompt Engineering
Function Calling
Semantic Search
Hybrid Search
Reranking
Time Series Forecasting
Anomaly Detection
Tokenization
Timeline
Ad-hoc Developer
•
Middle
Codingo Assignments
•
Freelance
Completed client coding tasks efficiently within strict project deadlines.
.NET Developer
•
Middle
Nawa Data Solutions
•
Contractor
Worked on Standard Chartered Bank's ongoing Loan Channeling System project.
Institut Sains dan Teknologi Terpadu Surabaya
Bachelor's Degree •
Computer Science
Web Developer Intern
•
Junior
Eyesimple Creative Studio
•
Internship
Worked on an ongoing online shop website development project.
All Tiers AI/ML Engineer
Confidence: Medium LLM Engineer
LLM-focused engineer at a Middle level who builds lightweight RAG and tool-integrated chatbot prototypes. The strongest proven skill is assembling end-to-end retrieval-augmented flows and tool-call integrations as shown by rag_chatbot.py (create_vector_db, rag) and calendar_booking_assistant.py (tool-call handling with add_event/update_event/delete_event). There is little evidence of rigorous ML training pipelines, experiment tracking, production-grade serving, or computational optimization in public code.
Model Architecture & Training
2/10
How well models are designed and trained
Basic model/agent logic present (Q-learning rules, reward/update functions) and lightweight use of embeddings; no custom neural architectures, optimizers or training pipelines.
Evidence
self-learning_tic-tac-toe-online/Self-Learning Tic-Tac-Toe (Online).py: Reward, UpdateQTable, GetState functions implementing a Q-learning agent
rag_chatbot/RAG Chatbot/rag_chatbot.py: create_vector_db uses OllamaEmbeddings(model='nomic-embed-text')
Data Pipeline & Feature Engineering
2/10
How data is prepared for models
Simple data acquisition and preparation for retrieval (web search -> text splitting -> vectorization); no large-scale ETL or feature engineering.
Evidence
rag_chatbot/RAG Chatbot/rag_chatbot.py: create_vector_db uses TavilySearchResults.invoke to fetch web content and RecursiveCharacterTextSplitter to create document chunks
rag_chatbot/RAG Chatbot/rag_chatbot.py: Chroma.from_documents(...) to build the vector store
Experimentation & Evaluation
1/10
How results are measured and tested
Minimal experimentation or evaluation artifacts; basic in-code reward/decision logic but no experiment tracking, held-out evaluation, or metrics pipelines.
Evidence
self-learning_tic-tac-toe-online/Self-Learning Tic-Tac-Toe (Online).py: Reward and Draw functions used for agent evaluation and Q updates
MLOps & Deployment
2/10
How models are shipped to production
Prototype deployment via Gradio interfaces and direct API/tool integration; lacks production serving, versioning, monitoring or CI/CD for models.
Evidence
rag_chatbot/RAG Chatbot/rag_chatbot.py: gr.Interface(...) and interface.launch(debug=True)
calendar_booking_assistant/Calendar Booking Assistant/calendar_booking_assistant.py: setup_google_calendar_api_service uses googleapiclient.build and service is used by tool-calling code
Computational Efficiency
1/10
How efficiently computing resources are used
No evidence of GPU/efficiency optimization, batching, quantization or profiling; only small-scale algorithmic work in Python.
Evidence
self-learning_tic-tac-toe-online/Self-Learning Tic-Tac-Toe (Online).py: Q table construction and single-threaded loop; no GPU/parallelization code
Research Depth & Innovation
1/10
Depth of research and new ideas
No novel research contributions or paper implementations; basic Q-learning and simple RAG composition but no innovation at research level.
Evidence
self-learning_tic-tac-toe-online/Self-Learning Tic-Tac-Toe (Online).py: basic Q-learning algorithm implementation
rag_chatbot/RAG Chatbot/rag_chatbot.py: straightforward RAG flow using web search, splitter, embeddings and Chroma
Expertise
AI Agents & Agentic Workflows• Middle
Recommendations
- Develop lightweight RAG prototypes and conversational assistants that integrate web-search, vectorization and tool-calls for rapid POCs.
- Build agentic integrations that call external APIs with robust input validation, retries/backoff, and structured tool-call handling.
- Harden existing prototypes for production: add authentication/secret management, error handling, observability and CI/CD.
- Expand ML engineering skills with experiment tracking and evaluation (W&B or MLflow) and small-scale efficiency work such as batching or quantization for inference.
Repositories
The developer's experience in this domain has been verified based on AI analysis of the following repositories: