# Divit Kashyap First Class Computer Science graduate from the University of Bristol, graduating on 8 July 2026. Based in London, freelancing and seeking software engineering roles. Divit builds end-to-end software across computer vision, full-stack applications, open-source developer tools, AI agentic workflows, and creative coding. He is especially interested in understanding every layer of the systems he works on: user experience, frontend engineering, backend architecture, infrastructure, data flow, performance, and developer workflows. AI-assisted development has made this style of building feel more natural: moving across layers quickly, checking assumptions, and thinking like a system designer while still shipping practical applications. Divit is a passionate builder, curious learner, and contributor who wants to make useful things with real impact. ## Focus Areas - Computer vision - Full-stack engineering - Open-source tools - AI agentic workflows and developer tooling - MCP skills and agent integrations - Creative coding and interactive web experiences - High-performance computing ## Experience ### Software Engineer Intern - Index Funds Sahi Hai Fintech internship focused on interactive web tools and fund analytics. Built tools for a platform managing over $1.2M in assets, including NAV comparison workflows, SIP/EMI wealth calculations, automation around content and internal stores, and low-latency financial data experiences. ### Graduate Teacher - University of Bristol Mentors a student team building a YOLO-based machine-learning auto-labelling pipeline. Supports code review, Agile project management, GitHub workflows, AI-assisted development workshops, technical decision-making, and programming lab support for students. ### Research Intern - University of Bristol Conducted thematic analysis of student experience across the Software Engineering Project unit and synthesised curriculum recommendations from survey and focus-group data. ### Software Tools Demonstrator - University of Bristol Provided debugging support, code review, and programming help across first- and second-year programming labs. ## Selected Projects ### MiniMax AI Vision Analysis Skill Contributor to a structured AI-agent skill for computer vision tasks including image description, OCR, UI review, chart extraction, and object detection. Installed thousands of times across multiple AI coding platforms. ### Loch A notch-native macOS control centre for media, clipboard history, AI-agent sessions, build watchers, dev ports, dictation, and daily operations. Built with Swift, AppKit, SwiftUI, and Homebrew distribution. ### Brick by Brick An AI-assisted, offline-first Flutter mobile app for daily planning, recurring habits, health-aware reminders, weekly commitments, widgets, Live Activities, Dynamic Island, and reflection. Led product strategy, UX direction, architecture decisions, testing, privacy planning, and release readiness across Flutter, Drift/SQLite, Swift/SwiftUI, HealthKit, EventKit, Health Connect, local notifications, WidgetKit, ActivityKit, App Intents, and on-device voice capture. ### Open-Wheel Racing Prototype A playable F1-style racing MVP built in Godot 4.6 with real-time driving, keyboard/controller input, lap and sector timing, multiple circuits, racing assists, setup tuning, AI rivals, persistent ghost laps, a broadcast-style HUD, ERS/DRS-style systems, trackside scenery, web-safe rendering profiles, GUT tests, headless smoke tests, browser QA, Python tooling, Git, and Git LFS. It is an unofficial non-commercial fan/prototype project with licence-aware handling required before any polished public release. ### Hallucinated Gambits A Python/Numba chess engine developed through 11 submission versions for AI Chessathon 2026, reaching peak competition rating 1704 and displayed rank 211/465. Directed AI-assisted development while comparing baselines, testing regressions, assessing actual losses, and rejecting unsupported neural or heuristic changes. The project demonstrates search optimisation, reversible move generation, transposition tables, move ordering, selective search, king-safety evaluation, adaptive clock management, opening preparation, ML evaluation literacy, and evidence-based release discipline. ### He(a)rd ETH Oxford 2026 Main Track winning opinion market where users bet on the majority view, not their own, to surface honest crowd-sourced data. Won $10,000 investment from HomeDAO. ### Cyclist Rear-Collision Alert Final-year computer vision project using YOLOv8n on Raspberry Pi for rear-facing cyclist safety, temporal risk scoring, and multimodal alerts. ### LBM Fluid Simulation on Isambard 3 High-performance D2Q9 Lattice Boltzmann simulation optimised for the Arm-based Isambard 3 supercomputer using C, OpenMP, MPI, SIMD, and cache-aware data layouts. ### Chameleon TUI Theme-aware animated colour post-processing plugin for the OpenCode TUI with many visual styles and configurable rendering effects. ### OpenCode Contributions Contributions to anomalyco/opencode across session lifecycle, scroll behaviour, shell PATH handling, image attachment triggers, SQLite resilience, and TUI reliability. ### Project Houdini Quantum optimisation project using QAOA, Qiskit, and NetworkX for graph-based security scenarios. Won 1st place at the University of Bristol Quantum Hackathon. ### Versioned Work-in-progress pose editor for choreographers exploring GitHub-style version control for movement, MediaPipe pose extraction, frame-level skeleton editing, and sequence persistence. ## Awards - ETH Oxford 2026 Main Track Winner: $10,000 investment from HomeDAO. - Best High Performance Computing Project Prize: awarded by NAG in a cohort of 70 students. - UoB Quantum Hackathon 2025: 1st Place. - ETH Oxford 2025 NerdO Prize: $1,000. - Sui Foundation CTF: Top 10, $500. ## Links - Website: https://divitkashyap.com/ - GitHub: https://github.com/divitkashyap - LinkedIn: https://www.linkedin.com/in/divit-kashyap-64626221b/ - X: https://x.com/divitkashyp