
Quicky Prime
TunisiePublication : 18 nov. 2026
[Project-09] - LLM Gateway & Conversation Intelligence
- Stage
- 4-6 mois
- Présentiel
- Échéance : 2 déc. 2026
- AI & LLM Ops
- Cost Optimization
- Network Analytics
Description du poste
Overview: Build a cost‑optimized LLM platform around an LLM gateway with model routing, semantic caching and usage/cost tracking to generate summaries, action items, sentiment and quality scores from conversations.
Objectives:
- Self‑host and benchmark open‑source models
- Implement routing, caching and cost tracking
- Build the conversation analysis pipeline
Key features:
- LLM gateway with model routing by task and cost
- Semantic cache to reuse similar answers
- Per‑request cost and usage tracking
- Conversation summaries and action items
- Sentiment analysis and quality scoring
Tech stack: vLLM/Ollama, Qdrant, Redis, FastAPI
Required application documents: Attach your CV (PDF). Optionally include a short cover letter, your GitHub/portfolio link and your university internship agreement.
Selection process: 1) Email with correct project code 2) Profile review 3) Technical interview 4) Final answer.