Javlon Baxtiyorov
← All projects
Production system  ·  in production  ·  2026 – present

AI Automation for Business Processes

Large products where the AI does the repetitive work, reliably.

Production AI automation built into large business products — turning slow, manual, repetitive processes into LLM pipelines that run unattended, stay observable and fail safely.

Role
Senior software engineer · Citifuel
Impact
Reliable, auditable automation
Status
In production
Year
2026 – present
Fig. 4 — Load-Bearing Structure: stone arch bearing 3,500,000 monthly users with voussoirs, keystone, and foundation 3,500,000 MONTHLY USERS KEYSTONE VOUSSOIR INTRADOS EXTRADOS SPAN RISE BACKEND R₁ R₂ r = 255 fig. 4 — load-bearing structure
fig. — ai automation for business processes

The problem

Most “AI features” are demos that break on real business data, volume and edge cases — with no answer for what happens when the model is wrong.

Most 'AI features' are demos that break the moment they meet real business data, volume, or edge cases. A business process can't be automated by a clever prompt alone — it needs retries, validation, audit trails, human-in-the-loop checkpoints, and a clear answer for what happens when the model is wrong. An automation that's right 95% of the time and silent about the other 5% is a liability, not a feature.

What I built

LLM and NLP services wrapped in async job pipelines (FastAPI, Celery, Redis) so the heavy generation never blocks a request. Each automation is a defined process — structured inputs, validated outputs, retries, and an audit record — with humans able to review or override at the points that matter. RAG and prompt pipelines keep the model grounded in the business's own data rather than guessing. The hard part of AI automation isn't the model — it's everything around it: the retries, the guardrails, the trail you can audit afterwards.

Key decisions

Treat the model as an unreliable worker that needs supervision, not a black box to trust. Make every automated step observable and reversible. The goal is not to remove people, but to remove the repetitive work that was never a good use of them — and to do it without quietly introducing errors at scale.

The result

Every automated step is validated, observable and reversible, with humans able to review or override at the points that matter.

Reliable, auditable automation

Built with

LLM / NLP pipelinesRAG · prompt engineeringAsync jobs · Celery · RedisHuman-in-the-loopValidation · audit trailsFastAPIBusiness-process automation

Lessons from the bench

  •  An automation that's right 95% of the time and silent about the other 5% is a liability, not a feature.
  •  The hard part of AI automation isn't the model — it's everything around it: the retries, the guardrails, the trail you can audit afterwards.