About

About

About Me

Hi, I’m Ayaan Sharif. I build AI systems that work across video, audio, and text — mostly large-scale multimodal pipelines and agentic workflows, mostly proofs of concept and experiments.

Professional Summary

Most of my work is building systems that can actually process a lot of media at once — not just one video, one file — and making that run efficiently on real cloud infrastructure (A100/L4 GPUs), not just in a notebook.

I spent a couple years as a Full Stack AI Engineer at Xfinite Global PLC (erosnow.com), building multimodal AI systems for content analysis.

🛠️ Technical Skills

Programming Languages

  • Python (Primary)
  • SQL for data management
  • JavaScript/TypeScript for full-stack development
  • Bash/Shell Scripting for automation

AI/ML Frameworks & Libraries

  • PyTorch, TensorFlow - Deep learning frameworks
  • Hugging Face Transformers - Pre-trained models
  • Scikit-learn - Machine learning
  • OpenCV - Computer vision
  • Pandas, NumPy - Data manipulation
  • LangGraph - Agentic AI workflows

Backend & Full-Stack Development

  • FastAPI - High-performance API development
  • React.js - Frontend development
  • REST APIs - API design and implementation
  • HTML/CSS - Web technologies

Cloud & DevOps

  • Google Cloud Platform (GCP) - Vertex AI
  • Docker - Containerization
  • Git - Version control
  • Linux (Debian) - System administration

Databases & Specialized Tools

  • Redis - Caching and session management
  • Weaviate - Vector database
  • ImageBind - Multimodal embeddings

Specialized Areas

  • Multimodal AI (Video, Audio, Text processing)
  • Retrieval-Augmented Generation (RAG)
  • Agentic AI Workflows
  • Large Language Models (Gemini)
  • High-Throughput Inference Pipelines
  • Video/Audio Processing (Whisper)

💼 Professional Experience

Full Stack AI Engineer

Xfinite Global PLC (erosnow.com) | Mumbai, India (Hybrid)
April 2024 – 2026

  • Developed and deployed large-scale multimodal AI systems analyzing video, audio, and text using Gemini, InternVLM, and ImageBind
  • Built automated pipelines for scene segmentation, face detection/tracking, event tagging, and compliance analysis at scale
  • Engineered high-throughput inference pipelines on GCP (A100/L4 GPUs) using FastAPI
  • Designed RAG-based workflows for dynamic content summarization, subtitle generation, and semantic movie indexing
  • Contributed to agentic AI pipelines for scalable content intelligence and enhanced user recommendations

🚀 Key Projects

TensorCraft - AI/ML Role Preparation Platform

February 2024 – Present · Live Site

An AI-powered platform for preparing aspiring AI/ML engineers for technical roles:

  • Architecting RAG pipelines leveraging Gemini models via Google AI APIs
  • Developed backend infrastructure using FastAPI and Redis
  • Built interactive frontend with React.js
  • Implementing AI agent workflows for hands-on tool learning and interview preparation

Bellabeat Wellness Data Analysis

October 2023 – December 2023

Data analysis case study on Fitbit usage patterns:

  • Performed exploratory data analysis on wellness data using Python, Pandas, Matplotlib
  • Identified key trends in user activity, sleep patterns, and health metrics
  • Created actionable insights for product improvements and marketing strategies

🎓 Education

B.Tech in AI & Data Science
Mumbai University | Mumbai, India
April 2020 – April 2024

🌟 Interests & Activities

  • Open Source Contributions: Active contributor to AI projects on GitHub and Hugging Face
  • Tech Community: Engaged in AI and data science discussions on Twitter(X) and Discord
  • Linux Enthusiast: Experienced Debian user passionate about system optimization
  • Continuous Learning: Always exploring the latest developments in AI and machine learning

📫 Let’s Connect!

Always happy to talk about AI projects or new tools people are using — feel free to reach out.

Open to work involving multimodal AI and agentic workflows.