Mamihery Randriambolatiana

- Lead the design and delivery of production-grade GenAI systems used as core components of internal and client-facing products. - Own AI initiatives end to end: functional framing with non-technical stakeholders, AI architecture design, and hands-on implementation. - Design and implement LLM-centric architectures including RAG pipelines, hybrid search (sparse + dense), reranking, and agent-based workflows. - Develop production AI systems using Python, FastAPI, LangChain, LangGraph, Redis, Qdrant, SPLADE, and custom reranking models. - Fine-tune and adapt generative models for domain-specific use cases (instruction tuning, PEFT / LoRA). - Provide technical leadership and supervision to AI and backend engineers. - Collaborate directly with CEOs, product teams, and non-technical stakeholders to translate business needs into scalable AI products. - Designed and built the full AI layer of myAuxilium, an enterprise knowledge enablement platform powering semantic search, document understanding, and knowledge retrieval in production.

- Led generative AI projects from conception to delivery. - Acted as the main technical and functional interface between stakeholders and AI teams. - Supervised development efforts around LLM-based systems. - Applied model optimization techniques, including quantization and Quantization-Aware Training (QAT), to improve inference efficiency. - Structured delivery processes bridging business expectations and technical execution.

- Developed machine learning models for intent detection, topic modeling, and named entity recognition. - Contributed to data pipelines and data warehouse implementation. - Automated analytical and operational workflows, including RPA use cases. - Participated in early AI initiatives that later evolved into production GenAI systems.