About Me
Hello! I'm Mohammed Ashraf
I'm a Machine Learning Engineer with over 6 years of experience specializing in Generative AI, LLMs, and Agentic AI. I'm passionate about building scalable, high-impact AI solutions, particularly in fast-paced startup environments. I have a strong background in developing and deploying real-time applications, RAG-powered chatbots, and voice agents. My expertise includes fine-tuning foundational models for domain-specific tasks and leading teams to deliver successful projects.
Download CVMy Skills
Machine Learning
Deep Learning, Model Training, Evaluation
Python
Expert in Python programming
Flask
Web application development
Prompt Engineering
Crafting effective prompts for LLMs
Hugging Face Transformers
Working with transformer models
LLMOps
Working with LLMOps
Natural Language Processing
NLP techniques and applications
FastAPI
Building high-performance APIs
Generative AI
Developing generative models
LangChain
Building applications with LLMs
AWS
Cloud computing services
AgentOps
Agent operations and management
JQuery
Interactive User Interface
Bootstrap
Interactive User Interface
MogoDB
NoSQL DataBase
My Projects
Odia T5-Based Model for NLP Tasks
Developed a T5-based model for NLP tasks in the Odia language, including question answering, summarization, and translation. Trained on the mT5 architecture.
View ProjectAgriQBot
Fine-tuned a FLAN-T5 model for question answering with a custom dataset, achieving a BLEU score of 0.86.
View ProjectFine-Tuned Gemma 2B for Python Code Generation
Fine-tuned the Google Gemma 2B Instruct model for precise Python code generation from natural language instructions.
View ProjectReal-Time AI Telephony Sales Agent
Developed an AI-powered real-time sales agent with Twilio integration, using Groq and Deepgram for efficient speech recognition and NLP.
View ProjectBuilding a Voice-Enabled AI Agent for Text-to-SQL Queries with LangGraph
This project bridges the gap between non-technical users and databases by building a voice-enabled AI agent that translates natural language (text and speech) into SQL queries
View ProjectCSV To Conversational Chat Bot
Chat-React-CSV-Bot is a sophisticated conversational agent engineered with OpenAI's GPT-3.5 model and React agent. This project integrates natural language processing capabilities to develop a chatbot adept at comprehending and generating responses to user inquiries.
View ProjectWork Experience
Senior Data Scientist
MResult - Bengaluru, India
Developed a Gen AI project management tool integrated with Jira, optimizing prompt engineering and automating document generation.
Senior Machine Learning Engineer
Kapture CX - Bengaluru, Karnataka, India
Engineered a multilingual chatbot, implemented an Agentic RAG KMS, and developed an NLQ dashboard, significantly improving support efficiency.
Chief Technology Officer
Sales Sunday – Gurugram, Haryana, India
Implemented AI-based QA and script generation, boosting lead conversion and reducing costs. Created a RAG-based KMS for real-time support.
Lead Machine Learning Engineer
Wobb - Gurugram, Haryana, India
Developed a multimodal classification model, implemented an influencer discovery algorithm, and designed a data pipeline for campaign analysis.
Co-Founder
Frejourn - Bangalore, Karnataka, India
Led backend development, implemented personalization algorithms, integrated APIs, and participated in Y Combinator and Newchip accelerators.
Blog's
Building a Retrieval-Augmented Generation (RAG) Model with Gemma and Langchain: A Step-by-Step Guide
Learn how to create powerful RAG models using Gemma and Langchain in easy steps! This guide helps you build advanced text generation systems effortlessly.
Read MoreFrom Manual to Automated: The Future of Web Scraping with LLM
In web scraping, the main challenge has always been the manual effort required.However, those days are behind us. With the advent of Large Language Models (LLMs) equipped with vision capabilities, we can now create nearly universal web scraping agents, significantly simplifying and automating the process.
Read MoreHow to Generate Structured Outputs with Google’s Gemini API: A Comprehensive Guide
This blog highlights the growing importance of generating structured JSON outputs in AI and introduces Google’s Gemini API as a versatile tool for creating structured data from text, images, or multimodal inputs.
Read MoreA Guide to Supervised Fine-Tuning and 4-Bit Quantization for Language Models, Pushed to the Hugging Face Model Hub
This blog post focuses on saving and pushing your fine-tuned model to the Hugging Face Model Hub, offering practical tips to overcome common challenges, especially with quantized models, ensuring easy accessibility for the community.
Read More