Avinash HM

    Avinash HM - AI Engineer

    Avinash HM is an ai engineer building production-ready AI systems, RAG pipelines, agentic workflows, and cloud-backed applications. Core strengths include LLMs, NLP, Agentic AI, Edge AI Optimization, ML Pipelines.

    Projects

    Sahayak AI

    Agentic AI platform designed to assist teachers with administrative tasks, lesson planning, and student engagement. Built for Google Cloud's Agentic AI Day 2025 hackathon.

    Stack: LangChain, Python, LLMs, Google Cloud, Genkit

    High Performance AI Video Gen

    Developed an optimized video generation engine leveraging Lightricks LTX-Video distilled models. Implements a 3D transformer architecture with rectified flow scheduling to produce high-fidelity, cinematic videos from text/images.

    Stack: Lightricks LTX-Video, Python, 3D Transformer, Distillation

    LLM Optimization & Quantization

    Deep dive into model compression (AWQ, GPTQ, GGUF) for LLaMA 3.2, Qwen-3, and GPT-2 to reduce memory footprint. Conducted rigorous benchmarking across factual vs. creative tasks using a custom evaluation framework.

    Stack: AWQ, GPTQ, GGUF, llama.cpp, PyTorch, LLM.int8()

    CalSnap

    Fitness tracking application empowering users to log meals with photos (instant calorie estimation), track progress metrics, and receive personalized adaptive workout/meal plans.

    Stack: React Native, AI Vision, Health Tracking

    RL_model_Snake_game

    Minimal modern reinforcement learning Snake game environment. Includes custom Gymnasium environment, PPO agent training, model saving/loading, and GUI visualization with TensorBoard logging.

    Stack: Gymnasium, Stable Baselines3, PyTorch, RL, Python

    LLM Finetuning using Unsloth

    Fine-tuned the Qwen3.5 0.8B model over a medical dataset to generate highly accurate and context-aware responses. Delivering correct and better-formed sentences specifically tailored for the healthcare sector.

    Stack: Unsloth, Qwen3.5, LLM Fine-tuning, Python

    CyberSync

    LinkedIn reconnaissance tool developed for red-team security research and awareness. Demonstrates ethical hacking practices and security analysis.

    Stack: Python, Web Scraping, Data Analysis, APIs

    Multipurpose Agriculture Robot

    Sensor-driven automation system for agricultural operations including monitoring, irrigation control, and crop management.

    Stack: Embedded Systems, Sensors, Automation, IoT

    Skills

    Core strengths span AI engineering, ML systems, cloud deployment, analytics, and cross-functional product delivery.

    AI & ML

    • Python
    • Machine Learning
    • Deep Learning
    • NLP
    • LLMs
    • Neural Networks
    • Transformers
    • RAG
    • Prompt Engineering
    • Fine-tuning
    • Model Quantization
    • Computer Vision

    Frameworks & Tools

    • LangChain
    • LangGraph
    • LangSmith
    • Hugging Face
    • Docker
    • n8n
    • Genkit
    • ADK
    • PyTorch
    • TensorFlow
    • OpenAI API

    Cloud & Deployment

    • Google Cloud
    • AWS Bedrock
    • Vercel
    • Cloud Functions

    Analytics

    • Power BI
    • Tableau
    • SPSS
    • MySQL
    • Data Analysis

    Soft Skills

    • Stakeholder Management
    • Communication
    • Leadership
    • Problem Solving
    • Team Collaboration

    Experience

    AI Engineer - Aigenthix Technologies Pvt Ltd.

    Current

    Developed AI chatbots for customer service and support. Built AI receptionist system for automated call handling. Created AI interviewer for candidate screening.

    ML Intern - Reliance Jio

    Previous

    Developed NLP-based email automation systems. Optimized edge LLM deployments for mobile devices. Implemented model quantization techniques.

    Independent International Business Consultant - X-Culture

    August 2024 - December 2024

    Competition analysis. Analysis of the international target market and trade regulations. Development of an international market expansion strategy (market entry mode, marketing, promotion, and product adaptation).