Aller au contenu
Toutes les missions
Veille marché · Hnrss Publiée il y a 3 jours

New comment by axegol in "Ask HN: Who wants to be hired? (November 2026)"

Location: Dubai, UAE (GMT+4) Remote: Yes (full-time or contract) Willing to relocate: Yes, with visa sponsorship Technologies: Python, TypeScript, vLLM, Ray, LiteLLM, 4-bit quantization (AWQ...

Entreprise ion
Mode Remote

Résumé

Location: Dubai, UAE (GMT+4) Remote: Yes (full-time or contract) Willing to relocate: Yes, with visa sponsorship Technologies: Python, TypeScript, vLLM, Ray, LiteLLM, 4-bit quantization (AWQ/GPTQ), speculative decoding, RAG and agentic retrieval, LLM fine-tuning, OpenTelemetry, Prometheus/Grafana, Temporal, Airflow, Claude Code, Codex, MCP Résumé/CV: https://www.linkedin.com/in/aleksandr-golovin-axeg0/ GitHub: https://github.com/AXEG0 Email: axegggl@gmail.comApplied AI leader who takes LLM products from prototype to production on hardware the client owns.Head of Applied AI at Rihal (Muscat, Oman). I built SIRAAJ, an on-prem LLM assistant for government and enterprise teams that can't send their data to outside APIs. About 10,000 people in Oman use it, and routine tasks that took about two hours with a general chat assistant now take about 30 minutes. I ran its serving myself, from CPU-only machines to H200 clusters with models up to about 600B parameters, and led a cross-functional team of up to 12.Before that I led data science at Hybrid.ai, a programmatic ad platform. I built its ML and the team behind it from scratch, moved bidding to deep neural models serving up to 100,000 requests per second, and shipped a model only after an A/B test proved its lift: a several-fold gain over the baseline and a multi-million-dollar revenue contribution. My team also shipped automatic banner generation, whose banners converted about 30% better than designer-made ones.Earlier: data scientist at the crypto exchanges YObit and LATOKEN, and four years as an options quant at BCS Global Markets.Looking for Head/Principal/Staff roles in applied AI or LLM infrastructure.