Dr. Farshid Pirahansiah CV

I can help companies by leveraging my deep expertise in AI, machine learning, and product analysis to drive strategic innovation and transform how businesses approach complex challenges. As a GenAI Product Analyst, I bring a comprehensive skill set that includes conducting data-driven user research, defining product goals, and driving the development of AI-based solutions aligned with business strategies. My proficiency in agile methodologies, user-centric design, and advanced analytics allows me to build and launch effective digital products, ensuring they are both technically sound and customer-focused. This approach supports an organization’s mission to lead in AI-driven advancements and deliver exceptional value in competitive markets.

I can help companies by leveraging my deep expertise in quantitative analytics and machine learning to drive impactful solutions across business verticals. In a role like Lead Quantitative Analytics Specialist, I bring extensive experience in building, evaluating, and implementing complex models, including those involving large-scale data and Generative AI applications. My skill set includes applying advanced mathematical theory and AI techniques, collaborating with cross-functional teams to optimize model deployment, and providing strategic insights for model performance and risk management. With a strong foundation in end-to-end machine learning pipelines and the ability to mentor junior data scientists, I can contribute significantly to advancing data-driven innovation and maintaining compliance in highly regulated environments.

My Products

My custom ChatGPT

My Social

Index

Camera Calibration Expertise

I am an expert in camera calibration, with several publications, applications, and solutions developed around this area. My work spans both single-camera and multi-camera systems, where I have experience with a range of calibration techniques, from simple fixed patterns (like chessboards) to more complex, automated calibration methods.

Expertise Areas:

Calibration Techniques:

I have worked with various industries such as robotics, IoT, medical technology, and industrial automation, providing robust calibration solutions that ensure accuracy across multiple environments and platforms.

https://lnkd.in/d97ypMcc

Computer Vision + LLM

AI Model Cost Calculator: Optimizing Costs for Computer Vision and Multimodal AI Solutions

My AI Model Cost Calculator with the latest pricing for GPT-4 Turbo, Google Gemini 1.5 Pro, and Claude 3 Opus! 💡 Calculate text and image processing costs easily with real-time estimates. Check it out! #AI #MachineLearning #CostCalculator #GPT4 #Claude #GoogleGemini

OpenCV 5 NuGet Package – Simplified Setup for Visual Studio

Get started with OpenCV 5 effortlessly! My custom NuGet packages allow you to integrate the static library into your Visual Studio 2019 or 2022 projects in just a few minutes.

Key Features:

Step-by-Step Tutorial:

Check out my detailed guide on how to install and set up your OpenCV project on Visual Studio 2022. Watch it here:
Install OpenCV on Visual Studio in Minutes

Source Code:

Download the complete project on GitHub.

#OpenCV5 #C++ #ComputerVision #VisualStudio

cvtest: Computer Vision Test

Unit Test, Integration Test, System Test, Acceptance Test for Computer Vision and Deep Learning

Do you want to test the output of your computer vision application, whether it’s video or images?
In computer vision development, there are no standard tests readily available, so I’ve created and shared a few.

Key Features:

Find more details in the README file on GitHub.

Download the Source Code:

GitHub Repository

static opencv make 200K to 18MB but no need DLL

object detection and tracking on Edge and cloud Multi Camera (Stereo Vision) Calibration for AR/VR headset (extended reality/mixed reality) 3D Image Processing with Deep Learning

To remove unwanted objects from an image using contour detection performed twice, start by reading the image and applying a threshold or edge detection method to create a binary image. Detect the initial contours using cv2.findContours, and filter these contours based on a chosen area threshold to keep only the larger objects. Draw and fill the smaller, unwanted contours using cv2.drawContours with a cv2.FILLED parameter to mask them from the image. Next, perform a second pass of contour detection to identify any remaining objects and repeat the filtering process. Finally, draw the filtered contours onto an output image using cv2.drawContours with a visible color for visualization. Display or save the final image to confirm that only the desired objects remain. This double-pass method ensures effective removal of nested or smaller objects by progressively refining the contours in the image.

LLM Developers: VSCode Extensions Farshid

VS Marketplace Link: VSCode Extensions Farshid

VSCode Extensions Farshid is a pack of essential tools designed for Computer Vision (CV), Machine Learning (ML), Large Language Models (LLM), and Personal Knowledge Management (PKM) projects.

Enhance your workflow with tools like:

🔗 Get it on the Marketplace: VSCode Extensions Farshid

💬 Feedback: Check out the GitHub repo for issues, feedback, or contributions.
Happy coding! 💻✨

Extension Details:

#VSCode #DeveloperTools #Programming #MachineLearning #DataScience

farshid/portfolio/projects/GUI_OpenCV/realtime_GUI_OpenCV.py

Excited to share my new Real-time OpenCV Function Tester built with PyQt5! 🎉

🔹 Apply OpenCV functions on images with a user-friendly GUI. 🔹 Safe code execution environment. 🔹 Undo functionality for easy experimentation. 🔹 Perfect for learning and prototyping in computer vision.

Check it out! 🖥️✨

#OpenCV #Python #PyQt5 #ComputerVision #Coding

farshid/portfolio/projects/GUI_OpenCV/realtime_GUI_OpenCV.py

Todo App Overview and Features

t.me/de_mini_app_bot @de_mini_app_bot Innovating Task Management with Local Storage-Powered Todo App | Full-Stack Developer | Telegram Bot & Mini App

Data Flow:

  1. Loading Data: Upon loading, the app retrieves todos from IndexedDB for persistence.
  2. Initializing App: Updates calendars and renders todos across all sections.
  3. User Interaction:
    • Adding Todos: Users can add todos to any section, which are saved with relevant date metadata.
    • Navigating Calendars: Changing the selectedDate updates the calendars and re-renders todos.
    • Managing Todos: Users can mark todos as done/undone or delete them. Changes are saved to IndexedDB.
  4. Saving Data: Any modifications trigger saveToIndexedDB(), ensuring data persistence.

Date Management:

Compatibility and Limitations:

Usage Instructions:

Conclusion:

This Todo App offers a simple, efficient way to manage tasks across different time frames with automatic data persistence. By leveraging IndexedDB, the app ensures data retention without manual saving. With cross-device compatibility, the app is accessible to a wide range of users.

Shortcuts:

Potential Enhancements:

Technical Features of the Todo App:

Technical Architecture:

#TaskManagement #AWSLambda #JavaScript #HTML #CSS #TelegramBot #OpenAI #WebDevelopment #CalendarIntegration #OfflineApp #ResponsiveDesign #ProductivityTools

@pirahansiahbot: AWS

ASK from My resume Farshid Pirahansiah Telegram miniapp:
Building and Deploying a Fine-Tuned ChatGPT Model for Telegram Integration Using AWS Lambda. Creating a Powerful AI-Driven Telegram Bot to Answer CV Queries: A Comprehensive Guide.

Project Overview:

Key Steps and Technologies:

Project Summary:

This project demonstrates how AI and serverless computing can be combined to create a responsive Telegram bot that can answer any question related to my CV. The bot, found at https://t.me/pirahansiahbot, showcases expertise in data preparation, model fine-tuning, and AWS Lambda deployment.

Hashtags:

#AIPoweredBot #TelegramBot #AWSLambda #OpenAI #Serverless #GPT4Mini #FineTuning #DataAugmentation #ModelOptimization #CVAssistant #ChatGPT

@image_processing_farshid_bot:

🚀 My new Image Processing Bot! 🤖🎨
📸 Send an image, apply advanced OpenCV functions like cv2.Canny(img), and get instant results! Perfect for quick edits and AI-driven enhancements.

💰 Support the bot via TonCoin or purchase stars.
Bot Link: https://t.me/image_processing_farshid_bot

Hashtags:

#AI #ImageProcessing #Python #OpenCV #TelegramBot

🚀 Just deployed a robust Telegram bot on AWS Lambda!
From setting up webhooks to handling image processing and integrating payments, this guide covers it all. Learn how to optimize your bot for serverless environments!
#AWS #TelegramBot #Serverless #AI #CloudComputing

Project Overview:

This guide provides a comprehensive overview of setting up, deploying, and troubleshooting a Telegram bot on AWS Lambda, including webhook configuration, Lambda layer creation, and common development issues.

@item2cook_bot: AWS

Transform your photos into beautiful pencil sketches instantly! Just send an image, and SketchBot will do the rest.
https://t.me/farshidpirahansiahbot

Key Features:

Technical Architecture:

This AI bot:

Related Methods:

•	Fine-tuning LLMs for specific business applications
•	Optimizing LLM workflows to enhance productivity
•	Developing LLM-based tools for customer support and engagement
•	Integrating LLMs into existing company processes to automate tasks
•	Utilizing LLMs for data analysis and predictive modeling

  1. Ultra Low-Power, Low-Processor Devices (e.g., Watch Microcontrollers, Basic Microcontrollers)

    • Focus on implementing only minimal machine learning models or Tiny LLMs. • Methods: TinyML, Quantization Techniques, Model Pruning, Edge Impulse Integration.

  2. Ultra Low-Power, Low-Processor Devices (Common Edge Devices like Raspberry Pi 5)

    • Deploying lightweight versions of LLMs optimized for edge inference. • Methods: ONNX Runtime, TensorFlow Lite, Model Distillation, EfficientNet Variants.

  3. Ultra Low-Power, Low-Processor Devices with RISC-V Architecture

    • Optimizing machine learning and LLMs for open-source, scalable RISC-V chipsets. • Methods: Custom Compiler Optimization (e.g., TVM), RISC-V-specific ML Frameworks, Edge AI SDKs.

  4. Ultra Low-Power, Low-Processor Devices Equipped with Nvidia Chips

    • Leveraging Nvidia’s GPU capabilities for edge AI models while maintaining energy efficiency. • Methods: NVIDIA Jetson Platform, CUDA Optimizations, TensorRT, DeepStream SDK.

These contributions can help optimize machine learning and LLM capabilities for edge devices across different hardware, enhancing their application while maintaining low power consumption and processing requirements.

As a consultant, I help companies build AI-integrated workplace ecosystems by offering strategic guidance, culture development, technology implementation, and tailored talent strategies. With extensive expertise, I leverage unique perspectives to approach challenges creatively and adapt to evolving business landscapes. My client-first approach ensures that each solution aligns with company needs while maintaining a strong commitment to ethical AI practices that meet legal and compliance standards.

I support businesses by conducting strategy and gap analyses, optimizing HR processes, and leading alignment workshops for leadership buy-in. I develop internal communication strategies, implement change management journeys, and strengthen employer branding. I also review workflows, automate processes, design AI chatbots, and integrate predictive HR technologies. Additionally, I provide AI education, 1:1 coaching, and strategic talent solutions to build an AI-ready workforce and redefine recruitment strategies.

Command Descriptions for less:

Linux (less x.cpp): • The command less x.cpp is used to view the content of the file x.cpp in a paginated way. The less utility allows you to scroll through the file and navigate efficiently.

Useful Key Bindings: • n: Search for the next occurrence of a pattern (after using / or ? to search). • p: In the context of less, p is not typically a standard key binding, but you may be referring to using ? for searching backward (which behaves like previous). • q: Quit the less command and return to the terminal prompt.

Mac (assuming using less command as well): • On macOS, the less command works similarly to Linux, with key bindings: • :n: Move to the next file in the input list. • :p: Move to the previous file in the input list. • q: Quit the viewer and return to the shell prompt.

Additional Tips:

•	/pattern: Search forward in the file for a specific pattern.
•	?pattern: Search backward in the file for a specific pattern.
•	space or f: Move forward one page.
•	b: Move backward one page.
•	G: Go to the end of the file.
•	g: Go to the beginning of the file.

These commands enhance navigation and make working with files in the terminal more efficient.