starRecruiting: mentor
GPU/PyTorch Mentorship for Nanopore RNA
I'm a student seeking a research-computing mentor experienced in NVIDIA GPUs, PyTorch, or bioinformatics. My science-fair project uses DeepChopper to preprocess public nanopore RNA-sequencing data, but CPU inference is too slow. I'm looking for guidance on a smal
asteriskmentor preferred attributes:

I'm seeking a mentor with hands-on experience in NVIDIA GPU computing, CUDA-enabled PyTorch, or high-performance scientific computing, ideally involving bioinformatics or long-read sequencing.

Experience with GPU inference benchmarking, memory and batch-size optimization, and reproducible computational workflows would be particularly helpful. Familiarity with DeepChopper is welcome but not required.

I'd especially appreciate someone willing to guide a high-school student through a small, time-limited GPU compatibility benchmark, help evaluate scientific equivalence, and advise on possible access to academic computing resources or an eligible institutional sponsor.

I'm open to remote mentorship and flexible scheduling.

Details

I'm a student in California working on an independent computational biology science-fair project.

My biological question is whether certain chimeric RNA transcripts become enriched during macrophage inflammation beyond what can be explained by parental-gene expression and sequencing artifacts.

The discovery dataset contains ten public nanopore RNA-sequencing libraries with approximately 53 million sequencing records. My planned analysis uses DeepChopper for artifact reduction before identifying candidate chimeric transcripts, with an independent dataset reserved for validation.

I've encountered a computational bottleneck. My original CPU workflow was projected to require several months, and a recent benchmark using generated data and DeepChopper's supported direct-FASTQ workflow achieved only approximately 1.17 reads per second during inference.

I'm hoping to connect with a mentor experienced in NVIDIA GPU computing, PyTorch inference, or high-performance bioinformatics workflows.

My immediate goals are to:

  1. Identify a suitable GPU-based DeepChopper execution configuration.
  2. Run a small, time-limited synthetic benchmark to evaluate performance and scientific compatibility.
  3. Understand whether supervised access to university or ACCESS computing resources might be possible.

I don't currently have a university affiliation or an ACCESS allocation, so I would also appreciate advice about eligibility and institutional sponsorship.

My submission deadline is November 13, 2026. Even a short mentorship engagement or referral to someone with the relevant expertise would be extremely helpful.

Thank you for considering my request.

Status

Recruiting

Mentee

Default profile picture
Other