- β‘ Status: Open to work / Available for hire. Ready to build full-stack AI systems, agentic architectures, and high-throughput backends. Let's talk!
- π Building full stack web and mobile apps embedded with practical AI
- π Master student in AI & Data Science at CADT
- π§ Focus: AI systems, OCR pipelines, and high-throughput backend services
- π¬ Ask me about FastAPI, NestJS, Flutter, Next.js, or PyTorch
- β‘ Fun fact: i know nth but nth doesn't know me :')
- π Graduate studies: Master in AI & Data Science at CADT
- π¬ Computer vision & OCR: Building synthetic datasets and fine-tuning TrOCR & YOLOv8 for Khmer document extraction
- π€ Agentic systems: Designing multi-agent harnesses, custom MCP servers, and automation pipelines
- π Shipping: Developing full stack web apps, Telegram bots, and automated scheduling workflows
π€ AI Telemetry & Token Metrics (via Token Monitor)
ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
β± Today's AI Compute : 53.06M tokens ($9.40)
π€ Lifetime Token Volume : 3.43 Billion tokens ($840.53 across 124 days)
β‘ Prompt Cache Efficiency : 75.9% cache hit (2.61B cache-read tokens)
π§ Primary Driver : Google Antigravity (IDE & CLI) with Gemini 3.8 Flash
Model & Agent Distribution:
Gemini 3.8 Flash 3.17B tokens βββββββββββββββββββββββββ 92.3 %
Gemini 3.7 / Local Models 0.26B tokens βββββββββββββββββββββββββ 7.7 %
Agentic Engineering Insights:
β’ High-Efficiency Prompter : Strict ponytail & YAGNI enforcement with prompt cache reuse
β’ Autonomous Execution : 20,302+ lifetime tool calls across IDE & CLI workflows
β’ Custom Agent Harness : Zero-slop skill rules and local MCP servers in daily pairing
β’ High Reasoning Density : 8.35M reasoning tokens generated
ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
- π€ Autonomous coding: Google Antigravity CLI as daily driver with Gemini 3.8 Flash for multi-step reasoning, test-driven iteration, and codebase audits.
- π οΈ Protocols & tools: Model Context Protocol (MCP) servers and browser agents for live database inspections, research verification, and file operations.
- π Orchestration & pipelines: OpenClaw, n8n webhooks, and Claude Code for event-driven agent workflows and scheduled tasks.
- ποΈ Applied vision & OCR: Fine-tuning TrOCR and YOLOv8 models on low-resource Khmer script with synthetic data pipelines.
I'm an Early π€
π Morning 25517 commits βββββββββββββββββββββββββ 18.35 %
π Daytime 53280 commits βββββββββββββββββββββββββ 38.32 %
π Evening 38283 commits βββββββββββββββββββββββββ 27.53 %
π Night 21955 commits βββββββββββββββββββββββββ 15.79 %
π I'm Most Productive on Thursday
Monday 15101 commits βββββββββββββββββββββββββ 10.86 %
Tuesday 18340 commits βββββββββββββββββββββββββ 13.19 %
Wednesday 15317 commits βββββββββββββββββββββββββ 11.02 %
Thursday 28259 commits βββββββββββββββββββββββββ 20.33 %
Friday 15263 commits βββββββββββββββββββββββββ 10.98 %
Saturday 26421 commits βββββββββββββββββββββββββ 19.00 %
Sunday 20334 commits βββββββββββββββββββββββββ 14.63 %
π This Week I Spent My Time On
ποΈ Time Zone: Asia/Phnom_Penh
π¬ Programming Languages:
No Activity Tracked This Week
π₯ Editors:
No Activity Tracked This Week
π» Operating System:
No Activity Tracked This Week
I Mostly Code in TypeScript
TypeScript 53 repos βββββββββββββββββββββββββ 43.80 %
Python 19 repos βββββββββββββββββββββββββ 15.70 %
Jupyter Notebook 14 repos βββββββββββββββββββββββββ 11.57 %
JavaScript 12 repos βββββββββββββββββββββββββ 09.92 %
Shell 1 repo βββββββββββββββββββββββββ 00.83 %
Last Updated on 16/10/2026 09:58:05 UTC


