About
TextYess is an AI-first CRM that turns conversations across on-site channels, WhatsApp, email, and voice into revenue. Its AI agents autonomously sell to, support, and retain customers for B2C brands.
Responsibilities
•Build the agentic runtime for proactive agents, including context building, tool and retrieval orchestration, scoring, and guardrails
•Own production components from day one and make decisions about signal versus noise
•Improve grounding by addressing retrieval, prompt, and evaluation issues
•Combine exact-match and semantic search in a single ranking system
•Improve multilingual response quality
•Profile requests, test prompt and model changes, and ship measurable latency and cost improvements
•Improve the golden-dataset harness so evaluations predict production quality
•Rank, score, and suppress proactive insights so agents interrupt customers only when warranted
Requirements
•At least 2 years of professional software development experience
•Experience shipping at least one LLM feature used by real users, such as a chatbot, RAG system, or agent; a weekend project does not qualify
•Practical knowledge of RAG, including embeddings, vector search, chunking, reranking, and their failure modes
•Ability to analyze production traces and identify whether issues stem from retrieval, prompts, or tool selection
•Strong Python and backend fundamentals
•Measure latency, cost, and quality rather than relying on guesses
Nice to have
•Experience with search infrastructure (Elastic, Algolia, Qdrant, Weaviate)
•Fine-tuning
•Graph search
•Knowledge of e-commerce operations
Conditions
No conditions stated in the posting.
