In medieval times, a population the so-called Rasikajy was settled on the northeastern coast of Madagascar. This population was in contact with the Indian Ocean trading network, as attested by the presence of imported goods from China, India or Persia in some tombs in the necropolis of Vohémar, excavated in the first half of the 20th century. This population remains under-studied and little is known about their material culture or their way of life. The presence of iron slag in northeastern Madagascar has long been documented in archaeological literature, but no in-depth study had been carried out. The aims of this doctoral project were therefore to describe and better understand these metallurgical remains and the associated technical tradition. Three excavation campaigns and four additional survey campaigns were carried out between 2017 and 2021, completely renewing our knowledge on this area of Madagascar. Approximately 150 slag heaps spread over twenty locations have been described, representing about 450 tons of slag. However, these sites are concentrated in the southern half of the study area, thus defining a spatially delimited metallurgical district. The metallurgical production could be dated by radiocarbon dating to between the 11th and 14th century CE. Before this period, the Rasikajy already knew and used iron tools, but they imported this metal via the Indian Ocean trade. The combined approach of fieldwork and laboratory work has enabled the reconstruction of metallurgical practices, despite the fragile and poorly preserved remains. The Rasikajy technical tradition used a furnace in the form of a simple elliptical bowl dug directly into the sandy substratum, without any clay lining. No clay superstructure could be identified either. However, small walls were built of loose sand, sometimes reinforced with a few stone blocks. A single tuyere, made of stone or clay, was set into this sandy wall and connected to bellows. Lateritic ores, in the form of ferruginous concretions with remarkably high iron content, were reduced in these small structures. When an excess of slag was produced for the capacity of the bassin, the slag was drained off through a small channel dug in the sand. This technical tradition is observed throughout the metallurgical district. However, each site is slightly different, which shows a local adaptation of the technique, probably depending on the availability of raw materials, e.g. to make the tuyere. An detailed chemical (XRF) and mineralogical (XRD, optical microscopy and SEM-EDS) study revealed a high variability in slag composition. This variability is partly due to a high contamination of sand, which is the construction material of the furnace. Mass balance calculations also show that iron production was irregular from one smelting operation to another. In some cases smelting operations even failed, producing no iron at all. The technique appears to be poorly controlled, although the structure of the furnace certainly does not allow for complete control of each smelting operation. However, the variability is such that we believe that the metallurgists were non-specialised workers, producing iron only occasionally when metal ran out. This is clearly not a mass production but rather a small-scale production with occasional smelting campaigns to meet local iron needs. Finally, a reflection on the origin of this unusual technical tradition has been carried out. As the Rasikajy were in contact with other populations in the Indian Ocean, a transfer of technology could have taken place. In this case, parallels of this technique would have been found elsewhere in the Indian Ocean. No similar technique has been documented so far in archaeological, historical or ethnographic sources. Hence, it is possible that the Rasikajy reinvented a basic smelting technique from smithing techniques.