← Toutes les offres

VSORA conçoit des processeurs et architectures de puces pour l’inférence d’intelligence artificielle dans les centres de données et les déploiements en périphérie. Ses solutions visent à réduire la latence, la consommation énergétique et le coût des requêtes d’IA.

Repères sur VSORA
Modèle
B2B
Siège
Meudon-la-Forêt
Domaine officiel
vsora.com
Création
2015 · 11 ans
Dernière levée
Late stage · 2026-07-01
Montant annoncé
35 000 000 $
Offres ouvertes
4

Investisseurs mentionnés

Détails de l’offre

La description complète publiée par VSORA.

Voir l’offre originale ↗

Next generation AI inference super-chip partitionning study in TSMC 5nm process Company presentation : VSORA is a French fabless semiconductor company delivering ultra-high-performance AI inference solutions for both data centers and edge deployments. Our proprietary architecture achieves exceptional implementation efficiency, ultra-low latency, and minimal power draw—dramatically cutting inference costs across any workload. Fully programmable and agnostic to both algorithms and host processors, our chips serve as versatile companion platforms.

A rich instruction set lets them seamlessly handle pure AI, pure DSP, or any hybrid of the two, all without burdening developers with extra complexity. To streamline development and shorten time-to-market, VSORA embraces industry standards: our toolchain is built on LLVM and supports common frameworks like ONNX and PyTorch, minimizing integration effort and customer cost.

After the tape-out of Horus chip (TSMC 5nm process) in the Jotunn SIP (from 2 to 8 chip per package), in October 2025, the next iteration of the chip is already on the way and requires evaluating how to split the design in a relevant way to achieve better quality of result and lower runtimes. Skills (mandatory, nice to have):

  • Languages : System Verilog, SystemC, Python, Tcl, bash, C++, cmake
  • STA basics (setup-hold, recovery-removal, Clock-domain crossing, …)
  • DFT basics (stuck at tests, test at speed, memory bist, …)
  • Low power ( level-shifting, power switching, isolation, UPF-1801 standard, ...)
  • Physical implementation basics (floorplanning, placement constraints, CTS concepts, ...) Internship content Length : 6 months Tasks :
  • Hands on with the implementation flow of an existing subblock 5M instances) (4 -5 months)
  • Study design architecture of the macro-block
  • Propose partitionning strategies
  • Study impacts on the following flow steps
  • Extending study on other macro blocks (optionnal)
  • Report redaction (0,5 month dedicated) Contacts : [email protected](Grenoble), [email protected](Grenoble), [email protected](Paris) Location : Grenoble office (internship location) : CEMOI A110 12 rue ampere 38000 Grenoble Paris office (2-3 travels): VSORA 15 Rue Jeanne Braconnier 92360 Meudon