Skip to content
View lukiod's full-sized avatar
🏠
Working from home
🏠
Working from home
  • idk what timezon i live in lol

Block or report lukiod

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.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

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

Report abuse
lukiod/README.md

Hey, I'm Mohak. I build agent systems that ship real code, not demos

I either type code manually or make it autonomous

Github LinkedIn

I run an agent pipeline that finds real bugs across tracked repos, fixes them, tests them, and ships the PR directly. No staging step, no rubber stamp. Most of what's below came out of that.

What I work on

  • Agent pipelines → source, fix, test, ship real code, no human in the loop until it's already merged
  • ML infra and debugger tooling → gdb/lldb pretty printers, inference optimization, CUDA scheduling
  • Trustworthy AI → federated learning, zero knowledge proofs, explainability

Merged PRs, last 30 days (auto updates, not hand edited)

Repo What it was
NVIDIA/TensorRT-Model-Connect fix(cli): unify kv-cache-size byte parser between CLI and benchmark
tenstorrent/whisper Do not let MXR disable pointer masking in M-mode
tenstorrent/tt-installer fix(installer): disable firmware updates by default in container mode
microsoft/apm fix(audit): allow local Claude-skill deps to omit apm.yml
NVIDIA/TensorRT-Model-Connect Parse family import scanning with ast instead of regex
NVIDIA/cudf-spark Remove inert InMemoryTableScanExec allowances in cache_test.py [databricks]
griddynamics/rosetta Remove dead _validate_topic function
DeusData/codebase-memory-mcp test: cover wrong JSON types in the MCP robustness layer
bkuan001/halo-record Fix chain verification propagating a stale self-declared hash after a break
HKUDS/Vibe-Trading fix(onboard): create .env.partial with owner only permissions
yzhao062/awesome-auditable-ai Standards: add the tamper-evident log and attestation primitives
diegosouzapw/OmniRoute chore(quality): add an RTL layout ratchet
usestrix/strix Create credential files with owner-only permissions
DeusData/codebase-memory-mcp fix(graph-ui): rank Accept-Language by q instead of substring match

Background

  • Backend and automation infrastructure for a data-driven marketing platform → real recurring revenue
  • Cerebras Fellowship → paper to deployable product on Cerebras hardware
  • Deep Learning Research Intern, IIT Roorkee → SegFormer and CFAN work, 82% mIoU
  • AI Intern, Golden Bird Healthcare → multimodal clinical documentation system, 82% Dice score
  • Zerobase Startup School (top 20, Korea) · Google for Startups x Antler

Pinned Loading

  1. cudamgr/cudamgr cudamgr/cudamgr Public

    A cross-platform CUDA version manager that simplifies installing, managing, and switching between different CUDA toolkit versions.

    Rust 13 1

  2. DeepFake-Detection DeepFake-Detection Public

    this is the repo of deep fake detection with resnext and lstm

    Jupyter Notebook 3 1

  3. MershLab/harness MershLab/harness Public

    The agent that grows with you

    Python