balu@tashkent:~$ status --availability
[ OPEN ] remote contract work: AI systems, legacy modernization, full-stack, bots
[ TZ   ] UTC+05:00, overlaps with Europe mornings and US early mornings

❯ cd ~/niche-engineering

Research-grade open source in hard-to-hire domains. Protocol first, reproducible results, every claim backed by a test.

code · benchmark · results	code · findings · protocol
code · route demo · benchmark	code · download .exe · docs

❯ cd ~/telegram-web-a

live demo · feature branches

$ git branch --list 'feature/*' feature/remind-me reminders on any message feature/message-card message as a PNG card feature/chat-insights story-style chat stats feature/storage-usage media cache breakdown feature/catch-up summary of unread messages feature/chat-calendar activity heatmap calendar Built into the open-source Telegram Web A client, each feature in its own branch, all of them running together in the live demo.

❯ cd ~/ai-systems

live demo · code · v1.2.0	live demo · code · v1.2.0
live demo · code · v1.2.0	live demo · code · v1.2.0

❯ cd ~/llmops

module	what it does	run
llmops-release-guard	CI release guard for AI services: health, budget, rollback and model policy checks	demo
ai-observability-kit	latency, token usage, model cost, success rate, error budget burn	demo
rag-deploy-blueprint	production-style RAG deploy: Docker, health checks, config validation, runbooks	demo

❯ cat ~/stack

❯ git log --oneline --focus

+ research-grade open source in niche domains: COBOL migration testing, quant validation, optimization
+ local-first AI desktop apps for Russian and Uzbek speaking users
+ new features for the Telegram Web A client
+ Telegram bots, Mini Apps and Next.js products for clients
+ evals, RAG, agents and release safety for every AI feature
- demos without tests

❯ ls ~/field-notes

file	about
prompt-engineering-playbook.md	prompt patterns, schemas, critic loops, product-ready AI workflows
llm-evaluation-checklist.md	quality checklist for testing AI features before shipping
rag-system-blueprint.md	retrieval, chunking, metadata, citations and monitoring
ai-devops-ci-cd.md	CI/CD for AI apps: prompt tests, eval smoke checks, release safety
digital-media-growth-system.md	AI, content strategy, analytics and tooling as one growth system

❯ watch -n 43200 ./contributions