Employment type: Full-time Format: On-Site Location: Tashkent Company: Logistics Salary: From 800$ (Based on experience)

About the Company We are building a real-time AI monitoring platform for the trucking industry. Our platform analyzes driver-dispatcher communications from Telegram recruiter contact using 7 independent AI agents in parallel, continuously updating the status of each truck in real time.

The system processes thousands of messages daily in EN/UZ/RU without translation, responding in 2-3 seconds. Stack: Python, FastAPI, AsyncIO, PostgreSQL, Redis, Google Gemini.

Your Responsibilities Optimizing existing prompts You will analyze and optimize prompts across 7 agents, reducing the alert false positive rate from the current to 5%. Every change is tested on real messages and measured with metrics.

Requirements Must Have — At least 6 months of production-level prompt engineering experience (portfolio required) — Understanding of GPT, Gemini, Claude differences: tokenization, context window, temperature — Experience extracting strict JSON schema responses from LLMs (few-shot, output parsing) — Metric-driven approach: prompt → test → measure → improve cycle — English B2+ (prompts are written in English) — Git: branch, commit, PR

Nice to Have — Knowledge of the trucking industry — Multilingual NLP experience: code-switching, dialects, CB radio jargon — Python: writing eval scripts, editing prompt files — Evaluation frameworks: promptfoo, DeepEval, LangSmith — Proficiency in Uzbek, Russian, and English ....

Success Metrics (KPI) — Alert false positive rate: reduce to 5% — Real issue detection coverage: 95%+ — JSON validity rate: 99.5%+ — Accuracy across all languages: 90%+ — Agent latency: must not exceed 2–3 seconds

Tech Stack — LLM: Google Gemini Flash (production) — Backend: Python 3.11+, FastAPI, AsyncIO — Database: PostgreSQL, Redis — Deployment: Docker, Cloudflare Tunnel — Monitoring: Custom admin dashboard, token/cost tracking

How to Apply Please include the following with your application: — Portfolio — prompts you have written or their results — Before/after metrics — if you have optimization case studies — Brief intro about yourself: your experience and any connection to trucking or logistics

Tg: recruiter contact

Note: We are looking for specialists who can demonstrate real metrics, not just a vibe check.