CSE (AI & ML) graduate building end-to-end AI systems — real-time computer vision, generative AI, on-device LLMs and voice agents.
Based in Hyderabad, India. Contact: affanasif422@gmail.com.
Muhammad Affan Asif (also known as Affan Asif) is an AI/ML & Computer Vision Engineer from Hyderabad, India, specialising in computer vision, generative AI, on-device LLMs and voice agents. He has shipped 21 projects and won 12 awards, including Smart India Hackathon 2024.
Languages: Python, Dart, JavaScript, SQL
ML / DL: PyTorch, TensorFlow, OpenCV, YOLOv8/v11, MediaPipe, Scikit-learn, HuggingFace
GenAI & LLMs: Gemini API, Stable Diffusion, ControlNet, ElevenLabs, TinyLLaMA, LoRA, RLHF
Retrieval & Agents: RAG (LlamaIndex), ChromaDB, LangChain, MCP, Groq API
Backend & APIs: Flask, FastAPI, WebSockets, Twilio, Supabase, SQLite, Deepgram
Frontend & Mobile: Flutter, Gradio, React
End-to-end classroom AI platform: YOLOv11 attendance, gesture-based board capture, OCR-to-notes pipeline, and "Manimator" — an AI explainer-video generator combining Manim animations with TTS.
Built with: YOLOv11, MediaPipe, Gemini, ElevenLabs, Manim, Flutter
Computer-vision analysis of classrooms — attendance and engagement insights from camera feeds, with a live web demo.
Built with: Computer Vision, Analytics
Sports analytics pipeline: YOLO detection + ByteTrack multi-object tracking + K-Means jersey clustering for team assignment. Computes player speed, distance, ball possession and pass success — camera motion compensated via optical flow.
Built with: YOLO, ByteTrack, K-Means, Optical Flow
AI-driven smart city infrastructure & safety monitoring: pothole detection, waste identification and human-animal conflict detection, with Twilio authority alerts and a Flutter app.
Built with: YOLO, Flask, Twilio, Flutter, Random Forest
Interactive YOLOv11 object-detection demo running in the browser — try detection on your own inputs.
Built with: YOLOv11, Object Detection
Vision-based analysis for post offices — monitoring and operational insights from camera feeds.
Built with: Computer Vision, Analytics
Frontend for a cafe analytics system — computer-vision insights for cafe operations.
Built with: Computer Vision, Frontend
Fully offline multimodal AI personal assistant in a Flutter app. TinyLLaMA 1.1B quantized 4-bit NF4 (4.1 GB → ~590 MB), LoRA fine-tuning on just 0.13% of parameters, RAG pipeline with LlamaIndex + ChromaDB, custom RLHF feedback loop, Stable Diffusion + ControlNet image generation and a WhatsApp automation agent via MCP + LangChain.
Built with: TinyLLaMA, RAG, RLHF, LoRA, MCP, Stable Diffusion, Flutter
An offline legal assistant that answers law-related queries using a local LLM with retrieval over legal documents — private, on-device legal help with no internet required.
Built with: LLM, RAG, Offline-first
A generative AI web app that weaves ideas into structured output — deployed and live on Render. Watch the demo videos to see it in action.
Built with: GenAI, Full-stack
AI-powered game generation — describe a game and let an LLM generate a playable build.
Built with: LLMs, Code Generation
A real-time conversational voice agent — speak naturally and get spoken responses. Deployed live on Railway.
Built with: STT, TTS, LLM, WebSockets
Full-duplex AI voice sales agent over live phone calls: Twilio streaming + Deepgram STT + ElevenLabs TTS with barge-in, ~80% conversion across live test calls, sub-700 ms latency, and a 4-state tone-adaptive dialogue system driven by real-time sentiment.
Built with: Twilio, Deepgram, ElevenLabs, Gemini, VADER
A personal finance mapping tool — visualise and track where your money goes. Built at Datanyx.
Built with: Finance, Data Viz
A tool for downloading YouTube videos — grab any video for offline use.
Built with: Python, JavaScript
A Chrome extension that exports your ChatGPT conversations to clean, shareable PDFs — published on the Chrome Web Store.
Built with: Chrome Extension, JavaScript
A drawing game where AI judges your sketches against a prompt — draw fast, score high.
Built with: AI, Canvas, Game
The classic arcade whack-a-mole, rebuilt for the browser.
Built with: JavaScript, Game
A word game you control with your blinks — computer vision meets casual gaming.
Built with: Computer Vision, Game
An explosive arcade game clone — live and playable on Render.
Built with: Game, Clone
The infamous Flappy Bird, cloned and playable in your browser.
Built with: JavaScript, Game, Clone
Full-time · Sep 2025 – Present · 11 mos
Leading a ~200 member chapter. Organised HackRevolution, a college hackathon with a ₹3,00,000 prize pool — built the registration platform handling ~250 team registrations and ~360 participants.
Full-time · Aug 2024 – Sep 2025 · 1 yr 2 mos
Led 30+ projects, hosted Agentic AI, Mind Meets Machine and TechNova; conducted a hands-on Computer Vision workshop.
Skills: Web Scraping, Flask
Full-time · Nov 2023 – Aug 2024 · 10 mos
Led development of six web projects for workshops and multiple ML initiatives.
Skills: OpenCV, TensorFlow
Internship · Jan 2023 – Jan 2024 · 1 yr 1 mo
Started in the CSI tech chapter — building web and ML projects for college events and workshops.
Skills: Postman API, ORM
Internship · Hybrid · Jun 2025 – May 2026 · 1 yr
Industry internship at Wipro, working on real-world software and AI engineering problems.
Internship · Jun 2023 – Aug 2024 · 1 yr 3 mos
Robotics & automation club — conducted a seminar on Arduino programming and embedded systems.
Skills: TensorFlow
Part-time · Jun 2017 – Jul 2020 · 3 yrs 2 mos
Editorial and publishing work for the school magazine and student publications.
Part-time · Jun 2018 – Jun 2020 · 2 yrs 1 mo
Competed in international space settlement design — went on to win the 15th Asian Space Settlement Design Competition and place 2nd at the Indian National level.