Gargeya Sharma
Founder @ Edudojo.ai
In love with AI and psychology, especially how people learn get evaluated work more efficiently.
A timeline of leadership in AI architecture, deep-tech research, and the foundational social calibration forged in high-pressure environments.
Leading the AI revolution in Evaluation & Education. Architecting advanced Socratic AI systems localized in English, Hindi, Japanese, and Spanish for a unified global market. Orchestrating an active pilot program serving 2,000+ students at Parishkar College (Autonomous), bridging technical innovation with scalable pedagogical design.
Evvolv
Co-founded and engineered a core multi-agent system, creating autonomous AI workers designed to automate complex business workflows and operations for Small and Medium Enterprises (SMEs).
RVS Consensus+
Built end-to-end data pipelines and real-time dashboards (Dash/Plotly) processing 10M+ rows of transaction data from SIX, Reuters, and 15+ global Tier-1 banks. Engineered a Python-powered automation toolset that saved 5+ hours daily and accelerated new product development by 20x.
Imperial College London
Consulted at the Flapping Wing MAV Lab. Developed a modeless, data-driven control approach for bionic flapping-wing micro-aerial vehicles, pushing the boundaries of applied physical AI methodologies.
Mitchells & Butlers / Compass Group
Worked at AllBarOne Leicester Square, one of London's busiest bars, while pursuing a Master's degree. Honed exceptional social calibration (EQ), high-pressure teamwork dynamics, and rapid problem-solving skills.
Sep 2022 – Sep 2023
Queen Mary University of London
London, UK
Jun 2018 – Aug 2022
University of Petroleum & Energy Studies
Dehradun, India
Hold-to-talk dictation for any app. Local Whisper or Gemini STT pastes at the cursor, with an optional LLM reply and a quiet glass HUD. Private by default on Windows, macOS, and Linux.
Android voice keyboard that talks to Groq, Gemini, and OpenRouter from the phone. Native IME microphone, optional floating control, and live streaming into the focused field. No desktop backend.
An autonomous agentic loop implementation that pits plan generation against a simulated adversary to verify, stress-test, and refine code execution trees.
An AI-powered digital clone and personal cognitive agent designed to ingest custom knowledge bases, mimic speech styles, and simulate reasoning paths.
The open-source code behind this premium, interactive Next.js digital CV, utilizing Framer Motion for micro-animations and Tailwind CSS.
An autonomous Model Context Protocol (MCP) server that scans codebases, evaluates documentation health, and auto-generates improvements using agentic LLM workflows.
Automated data-cleaning submission for the OpenEnv Hackathon organized by Scaler School, Meta, and Hugging Face.
Academic research and codebase from QMUL (Distinction). Focused on Deep Reinforcement Learning & Computer Vision control architectures.
Technical writing series featuring 17+ published articles covering Denoising Autoencoders, YOLO Object Detection, and neural network implementation details.