Adityo Nugroho | AI/ML Engineering
Live Deployment
Data Records
Integrated Suites
Event-driven log analysis system processing high-velocity log streams with tumbling window batching. Features async architecture with cascading AI triage for critical event detection and semantic root cause clustering.
RAG-powered autonomous runbook with hallucination prevention. Maps specific error codes to exact repair procedures from technical manuals, eliminating hallucination risk with hybrid search and strict context enforcement.
Agentic AI with human-in-the-loop safety guardrails. Translates natural language operational commands into safe, deterministic Python function calls with mandatory approval gates preventing unauthorized executions.
Multi-step GenAI workflow for architectural design on Google Cloud. Orchestrated pipeline transforming text descriptions into photorealistic renders through three stages: text enhancement, sketch generation, and rendering.
Deterministic multi-table telecom data generator with causal ordering. Produces realistic telecom tables with bit-exact reproducibility via cascading seed propagation and Parquet schema validation.
End-to-end analytics pipeline from EDA to strategic insights. Comprehensive data science practice with statistical rigor, SHAP interpretability, and high-performance ML models identifying congestion as primary QoE driver (Cohen's d = -2.12).
Production-ready framework for telecom AI/ML solutions. Covers 6 telecom ML use cases with complete technical specifications, domain-informed data generators, and unified technical standards.
High-performance async trading bot with dynamic trailing take-profit. Features regime detection, Donchian channel gating, and Ed25519 authentication. Deployed as systemd service for 24/7 autonomous operation.
FIX protocol connector for Binance with defensive message parsing. Three FIX sessions (Market Data, Order Entry, Drop Copy) with ED25519 authentication and spread market making strategy.
Comprehensive market analytics using Binance public REST API. Multi-endpoint integration with dataclass-based parsing, connection pooling, and multi-timeframe analysis for complete market snapshots.
On-chain token holder analytics and whale tracking. Three-mode architecture with pagination handling, known label identification, and snapshot comparison for tracking holder changes over time.
Multi-chain DEX pair scanner and early token detector. Real-time token discovery with connection pooling and multi-chain support across Ethereum, BNB Chain, Polygon, and Base.