Skip to content
View DhruvSrikanth's full-sized avatar
:electron:
Thinking
:electron:
Thinking

Organizations

@autonlab @Center-for-Applied-AI @WecoAI

Block or report DhruvSrikanth

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Please don't include any personal information such as legal names or email addresses. Maximum 100 characters, markdown supported. This note will be visible to only you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
DhruvSrikanth/README.md

I'm a Founding Engineer at Weco AI, building self-improving AI agents for engineers and scientists (such as aideml, AIDE). My work has been used by OpenAI (MLE-Bench, GPT-4.5, o1, o3-mini), Meta (The Automated LLM Speedrunning Benchmark: Reproducing NanoGPT Improvements, AI Research Agents for Machine Learning: Search, Exploration, and Generalization in MLE-bench), Google (MLE-Star), Microsoft (R&D-Agent), Amazon (MLZero), Sakana AI (AI Scientist-v2), METR (RE-Bench, Measuring AI Ability to Complete Long Tasks), UK AI Security Institute (Inspect AI), CMU (TimeSeriesGym, CodePDE), Stanford (MLE-Dojo), UC Berkeley (IDA-Bench) as well as tranditional media (1, 2 & 3).

Prior to this, I was a researcher at the Robotics Institute at Carnegie Mellon University working on large multimodal models, robotics and time-series for healthcare (such as AutonFeat, BIFURC). I also spent time at TTIC and UChicago Booth working on fairness and alignment for vision models (such as ASAC Guided Curriculum Learning).

I enjoy building high-performance implementations of my work and applying AI to healthcare, climate, scientific problems. Outside of research, I enjoy spending time outdoors hiking, surfing, and playing golf.

   Website    GoogleScholar    Twitter URL   

Open Source & Publications

🧠 Deep Learning & AI

🏥 Robotics & Healthcare

  • Bifurcation identification for ultrasound-driven robotic cannulation (Oral) [paper]

⚡ High-Performance Computing

  • HuggingFace Text Generation Inference [code] [readme] [docs]
  • CUDANN (Distributed Deep Learning Framework accelerated on NVIDIA GPUs) [code] [readme]
  • FastConv (Multithreaded GPU CUDA kernels for Convolution) [code]
  • Ray Tracing (using Monte-Carlo Algorithm on GPU Accelerators) [code] [readme]
  • Advection Diffusion Simulation (Multithreaded, Shared Memory Parallelism, Distributed Memory Parallelism) [code] [readme]
  • N Body Particle Simulation (Multithreaded, Shared Memory Parallelism, Distributed Memory Parallelism) [code] [readme]
  • Distributed Poisson Solver (using Conjugate Gradient Optimization) [code] [readme]
  • Image Editor via Convolutions (using Work Stealing and Work Balancing parallel schemes) [code] [readme]
  • Sparse Linear Solver (using the Map Reduce parallelism scheme with Conjugate Gradient Optimization) [code] [readme]
  • Image Editor via Convolutions (using Pipeline and BSP parallel schemes) [code] [readme]
  • Twitter Application (using Producer-Consumer parallelism model) [code] [readme]

🤖 Machine Learning

  • Boosting and Boostability (An understanding through Computational Learning Theory) [code] [paper]
  • Financial Forecasting of TSLA (using sentiment analysis, feature engineering, and RNNs) [code] [readme] [report]

💻 Computer Systems

🎮 Games

  • Climbing Mount Everest (A text-based adventure game) [code] [readme]

📚 Other

Pinned Loading

  1. pytorch/pytorch pytorch/pytorch Public

    Tensors and Dynamic neural networks in Python with strong GPU acceleration

    Python 91.8k 24.8k

  2. huggingface/text-generation-inference huggingface/text-generation-inference Public

    Large Language Model Text Generation Inference

    Python 10.4k 1.2k

  3. WecoAI/aideml WecoAI/aideml Public

    AIDE: AI-Driven Exploration in the Space of Code. The machine Learning engineering agent that automates AI R&D.

    Python 963 139

  4. GoLLUM GoLLUM Public

    A compiler for GoLite, a simple mix between Go and C/C++. The compiler uses LLVM for its IR representation and is designed for an ARM64 backend architecture.

    Go 5 1

  5. blue-yonder/tsfresh blue-yonder/tsfresh Public

    Automatic extraction of relevant features from time series:

    Jupyter Notebook 8.9k 1.2k

  6. autonlab/AutonFeat autonlab/AutonFeat Public

    A High Performance Library for Time-Series Featurization.

    Python 25 1