Behzad Samadi bsamadi

Resume current task

cat .rtts/tasks.md | grep "In Progress" -A 20

Check what's next

cat .rtts/tasks.md | grep "Next (Prioritized" -A 15

Start new task

git checkout -b feature/REQ-ID-description

Commit format

Run Nemotron-Nano-12B-v2-VL with vLLM + pixi + Python Web UI

0) Prerequisites

NVIDIA GPU (>=24 GB VRAM recommended)
Linux OS (WSL also works)
Installed: pixi from prefix.dev

DeepSeek R1 Distill: Complete Tutorial for Deployment & Fine-Tuning

This guide shows how to deploy an uncensored DeepSeek R1 Distill model to Google Cloud Run with GPU support and how to perform a basic, functional fine-tuning process. The tutorial is split into:

Environment Setup
FastAPI Inference Server
Docker Configuration
Google Cloud Run Deployment
Fine-Tuning Pipeline (Cold Start, Reasoning RL, Data Collection, Final RL Phase)

Role type: ML Tech Lead (non-people management) Work location: Remote
Position overview:
- Play a critical role in shaping the strategy, defining metrics and counter-metrics, and setting goals and priorities for the team ML system design, ML excellence, and identify and develop need new metrics plan.
- Lead projects and teams accordingly including develop custom/novel ML architectures.
- Function as go-to person to escalate the most complex online / production performance and evaluation issues beyond modeling and how the machine learning system interacts with the other systems around it.
- Re-evaluate the tradeoffs of already shipped features/ML systems and drive large efforts across multiple teams.
- Significantly level up the quality of ML work in the broader team around you
- Develop highly scalable classifiers and tools leveraging machine learning, data regression, and rules-based models.
- Code deliverables in tandem with the engineering team
Adapt standard machine learning methods

	from pymavlink import mavutil

	# Set the connection string to your autopilot. This can be a serial port or a network connection.
	connection_string = 'udp:127.0.0.1:5800' # Replace with the appropriate connection details

	# Create a MAVLink connection
	master = mavutil.mavlink_connection(connection_string)

	while True:
	msg = master.recv_match(type='COMMAND_LONG', blocking=True)

	This game pad has been around quite a while, so I thought I would try to use it on the latest version of Windows 11, here is what I found (08/01/2023).

	Windows 11 doesn't recognise this device when you plug it in (no Plug'n'Play).
	It will work quite well eventually, but I found I needed to do the following.

	Before you start, make sure the F710 is in 'X' mode NOT 'D' mode (little switch on the front)

	1. Download the Logitech F710 Connect Utility:
	https://download01.logi.com/web/ftp/pub/techsupport/joystick/F710ConnectUtility_2.15.264.exe
	2. Use this utility to get the F710 installed

	sudo apt update
	sudo apt install nvidia-cuda-toolkit
	ls /dev/nvidia*
	nvcc -V
	nvidia-smi

	vbus_voltage = 19.461620330810547 (float)
	serial_number = 2060358F524B (int)
	hw_version_major = 3 (int)
	hw_version_minor = 6 (int)
	hw_version_variant = 24 (int)
	fw_version_major = 0 (int)
	fw_version_minor = 4 (int)
	fw_version_revision = 11 (int)
	fw_version_unreleased = 0 (int)
	user_config_loaded = False (bool)

	ODrive control utility v0.4.11
	/usr/local/lib/python3.6/dist-packages/fibre/shell.py:104: RuntimeWarning: coroutine 'InteractiveShell.run_code' was never awaited
	console.runcode('import sys')
	/usr/local/lib/python3.6/dist-packages/fibre/shell.py:105: RuntimeWarning: coroutine 'InteractiveShell.run_code' was never awaited
	console.runcode('superexcepthook = sys.excepthook')
	/usr/local/lib/python3.6/dist-packages/fibre/shell.py:106: RuntimeWarning: coroutine 'InteractiveShell.run_code' was never awaited
	console.runcode('def newexcepthook(ex_class,ex,trace):\n'
	/usr/local/lib/python3.6/dist-packages/fibre/shell.py:109: RuntimeWarning: coroutine 'InteractiveShell.run_code' was never awaited
	console.runcode('sys.excepthook=newexcepthook')
	Please connect your ODrive.