Ramkishan Rohila

Architected and deployed a production LLM-based customer analytics system serving live users, improving personalized offer relevance by 22% using profit-aware recommendation logic. Built end-to-end conversational AI platforms for WhatsApp and Web, supporting persistent multi-turn sessions, role-based knowledge access, and fault-tolerant message orchestration for production workloads. Designed scalable document ingestion and RAG pipelines using OCR, MongoDB indexing, and Redis caching, achieving low-latency responses under concurrent user traffic. Monitored production systems, evaluated prompt and pipeline iterations, and collaborated with product teams to improve reliability and response quality.

Engineered a production AI scheduling assistant handling 1k+ monthly bookings via automated voice, SMS, and call workflows, reducing manual coordination effort by 40%. Built a high-throughput document summarization service using FastAPI and LLMs, processing 10k+ records/hour and improving editorial turnaround time by 60%. Developed and optimized a YOLO-based medical imaging pipeline, improving X-ray implant detection accuracy by 30% across 1k+ real-world cases. Automated internal recruitment workflows using AI-driven resume analysis and interview simulation, reducing recruiter review time by 50%. Implemented CI/CD pipelines with Docker, GitHub Actions, and AWS, enabling weekly reliable deployments while sustaining 99.9% service uptime.

Integrated LLM-powered workflows with telephony systems to build a voice-based scheduling assistant, reducing average appointment booking time by 30% during pilot deployments.