Hi, my name is

Adityo Nugroho.

AI/ML Engineer - Telecom/Industrial Domain Expert | 18+ Years Network Optimization Leadership (Deputy GM at Huawei)

I build the AI tools my teams needed: event-driven log analyzers (63x noise reduction), RAG troubleshooting engines with hallucination-resistant guardrails, and agentic automation with safety gates.

GitHub LinkedIn Email

01. About Me

After 18 years leading network optimization for operators serving millions (Deputy GM at Huawei), I witnessed what breaks at scale and built AI systems to fix it. I pivoted into AI/ML engineering to address these bottlenecks, architecting and deploying 12 production-ready systems in 14 months (Nov 2024-Jan 2026), including AI/ML systems spanning GenAI on Cloud Run (CI/CD), RAG, agentic AI, MLOps pipelines, and production infrastructure.

I bring 18+ years of domain authority in leading network optimization for large-scale systems to the systems I now automate. My designs prioritize operational resilience, ensuring systems gracefully handle real-world constraints like partial outages, race conditions, and rate limits.

"I translate technical complexity into measurable business impact. I design for reliability under stress, not just functionality in demos."

02. Featured Projects

Gemini 2.0 • Async Python • RAG • Function Calling

TRINITY Operations Suite

AI-Powered Network Operations: Observe → Decide → Act

A cohesive production-ready suite solving the "Mean Time to Recovery" bottleneck with three integrated components:

  • Incident Commander: Async log analyzer processing 500 logs/sec with 63x noise reduction
  • NOC-Oracle: RAG-powered troubleshooting with strict citation enforcement and hallucination-resistant guardrails
  • Net-Ops Agent: Agentic AI with 100% human-in-the-loop safety gates
Streamlit • Docker • Cloud Run • Gemini 2.0/2.5

AI Studio ● LIVE

Photorealistic Interior Design: Orchestrated GenAI on Google Cloud Run

An orchestrated GenAI pipeline transforming architectural sketches into photorealistic renders. Features a deterministic 3-step workflow (Text → Sketch → Render) with image-to-image refinement for 1:1 structural alignment.

MLOps • XGBoost • Parquet • Reproducible Pipelines

Telecom Digital Twin + MLOps Pipeline

End-to-End MLOps: Synthetic Generation → Model Training → Batch Prediction

A complete MLOps ecosystem combining physics-based simulation with production ML workflows.

  • Digital Twin: Deterministic data generator producing 5.6M sessions with SINR/latency modeling. 100% reproducibility via cascading seed propagation. Parquet schemas with automated validation (FK checks, temporal validity).
  • QoE Analytics: End-to-end ML pipeline with XGBoost/LightGBM regression and classification models. SHAP interpretability identified congestion as primary QoE driver (Cohen's d = -2.75). Modular architecture with CLI automation and batch prediction.
Async Python • WebSockets • Systemd • Ed25519

Trailing Edge

Production Algorithmic Trading Bot

Proven 24/7 reliability on Cloud VPS during active market periods. Features proprietary dynamic trailing take-profit with exponential decay and regime-aware auto-compounding. Achieved sub-100ms latency with automated systemd-based orchestration. Demonstrates ability to ship and maintain always-on infrastructure.

View Complete Portfolio

03. Technical Stack

Languages & Frameworks

Python 3.10+, Python Async (asyncio), Streamlit, FastAPI

AI/ML Models

Google Gemini 2.0 (Flash/Lite/Pro), Vertex AI, Multimodal AI (Text/Image/Video)

AI Engineering

RAG, LangChain, Agentic Patterns, Function Calling, Google Gen AI SDK, Prompt Engineering

ML & Data Science

XGBoost, LightGBM, scikit-learn, SHAP, Pandas, NumPy, Synthetic Data (SDV), Text Embeddings

MLOps & Data Engineering

Reproducible Pipelines, Schema Validation, Model Evaluation, Batch Prediction, Parquet, CLI Automation, SQL

Cloud & Infrastructure

Google Cloud Run, Docker, Artifact Registry, Linux VPS, Systemd, CI/CD

Protocols & APIs

REST, WebSockets, FIX 4.4, JSON-RPC, GraphQL

DevOps & Tools

Git/GitHub, Pytest, Ruff, UV Manager

04. What's Next?

I am currently open for opportunities!

Available for AI/ML Engineering roles focusing on:

  • Telecom/Industrial AI: Domain-specific AI/ML solutions leveraging network optimization expertise
  • Observability & APM: Automated remediation and log analysis
  • Fintech & Trading: Low-latency decision systems
Get in Touch