Use Compute
Craftly Cloud provides researchers, developers, university faculties, and startup teams in Bangladesh with high-performance, cost-effective distributed compute power without requiring foreign currency credit cards or complex infrastructure management.
Who Can Access Compute?
- Scientific Researchers: Academic teams working on life sciences, molecular dynamics, epidemiology, and climate modeling.
- AI Developers & Startups: Teams fine-tuning open-source LLMs, building computer vision pipelines, or running high-throughput inference.
- Undergraduate & Graduate Students: Students working on thesis projects, engineering simulations, or machine learning competitions.
Supported Workloads & Frameworks
Craftly Cloud’s distributed execution layer supports standard containerized and micro-kernel workloads across the AI and scientific computing ecosystem:
- Deep Learning: PyTorch, TensorFlow, JAX, ONNX Runtime, and Hugging Face transformers.
- LLM Inference & Fine-Tuning: vLLM, llama.cpp, LoRA adapters, and multi-node batch inference.
- Scientific Computing: NumPy, SciPy, BLAST/Clustal bio-informatics, Monte Carlo simulations, and ray-tracing pipelines.
API & SDK Integration
Submitting a job to Craftly Cloud is as simple as making an API call using our Python or TypeScript SDK:
from craftly import CloudClient
client = CloudClient(api_key="your_api_key")
job = client.submit(
model="meta-llama/Llama-3-8B-Instruct",
task="batch_inference",
inputs="data/queries.jsonl",
min_nodes=10,
)
print(f"Dispatched job {job.id} across Bangladesh mesh")University & Researcher Grants
Under our pilot initiative with the Ministry of Science and Technology, Craftly allocates dedicated compute pools to verified researchers and university teams across Bangladesh at zero cost. To apply for a research compute grant, contact our academic partnerships team with your project proposal.