LogoFresh Jobs Ahead
Baseten
Baseten
11 – 50 Angajați
Sponsor pentru viza H1BInteligență artificialăÎntreprindereSaaS
Baseten oferă infrastructură pentru inferența modelelor destinată companiilor care implementează aplicații de învățare automată și inteligență artificială. Platforma sa este concepută pentru servirea rapidă și scalabilă a modelelor, combinând inferența cu debit mare, implementarea rapidă, scalarea automată, servirea securizată a modelelor pentru companii și suportul pentru împachetarea modelelor open-source. Baseten ajută echipele de inginerie și învățare automată să gestioneze cerințele operaționale ale infrastructurii pentru modele, astfel încât să se poată concentra pe dezvoltarea unor modele specifice domeniului. Compania activează în domeniile inteligenței artificiale, software-ului ca serviciu și tehnologiei pentru întreprinderi și are o echipă de 11–50 de angajați.
LinkedIn
Locurile de muncă ale companiei

Inference Performance Software Engineer at Baseten (Hybrid)

Baseten is hiring an inference performance engineer to optimize LLM runtimes, GPU kernels, and model-serving systems. The role focuses on improving production inference speed, hardware utilization, and cost efficiency.

USStatele Unite ale AmericiiCACaliforniaNYNew YorkWAWashington
180.000 $ – 360.000 $ / an
HibridNormă întreagăNivel mediuSeniorLimbă Obligatorie: englezăInginer full-stackInginer software
C++PythonPyTorch

Descriere

  • Productionize advanced inference methods such as quantization, speculative decoding, KV-cache reuse, chunked prefill, LoRA, guided generation, and custom scheduling or routing.
  • Analyze the full inference pipeline, optimizing kernel overhead, memory placement, request scheduling, prefill/decode separation, and cache-aware routing.
  • Increase tokens per GPU-hour and utilization while balancing latency, throughput, and cost.
  • Deploy and optimize emerging model architectures on new hardware platforms.
  • Create benchmarks spanning model types, batch sizes, sequence lengths, and hardware setups.
  • Submit improvements to open-source inference projects including vLLM, SGLang, and TensorRT-LLM.
  • Work with model, infrastructure, and customer-facing teams to deliver measurable performance gains.

Cerințe

  • Degree at the bachelor's, master's, or doctoral level in computer science, engineering, mathematics, or a related discipline.
  • Professional experience with at least one general-purpose programming language, including Python or C++.
  • Working knowledge of LLM optimization methods such as quantization, speculative decoding, and continuous batching.
  • Strong command of machine-learning libraries, particularly PyTorch, TensorRT, or TensorRT-LLM.
  • Demonstrated interest in and practical experience with large language models.
  • Thorough understanding of GPU architecture.

Beneficii

  • Competitive pay package with meaningful equity participation.
  • For U.S. employees, full medical, dental, and vision insurance coverage extends to employees and their dependents.
  • Flexible paid time off, plus a company-wide winter break.
  • Paid parental leave.
  • Fertility and family-building support through a Carrot stipend.
  • For U.S. employees, a company-facilitated 401(k) plan.
  • Opportunities to work with a range of machine-learning startups, supporting substantial learning and professional networking.

Locuri de muncă similare

Kreato Global | BPO and Language Solutions

Remote English-Spanish OPI/VRI Interpreter

Kreato Global | BPO and Language Solutions
201 – 500 Angajați
Asistență medicalăLogisticăOspitalitate
ARArgentinaBOBoliviaBRBraziliaCLChileCOColumbiaCRCosta RicaCUCubaDORepublica DominicanăECEcuadorGTGuatemalaHNHondurasHTHaitiMXMexicNINicaraguaPAPanamaPEPeruPRPuerto RicoPYParaguaySVEl SalvadorUYUruguayVEVenezuelaRemoteNormă întreagăJuniorNu este necesară o diplomăLimbă Obligatorie: englezăLimbă Obligatorie: spaniolăTraducător

Interpret remotely between English- and Spanish-speaking people in medical, financial, social service, and customer care settings. Provide language support for Kreato Global across Latin America.

Deschide