Products
A cross-section of what we’ve built for teams — from vision-in-the-loop systems to consumer marketplaces to LLM-powered document tools. These are patterns we can build for you too.
Student-facing exam engine covering 11+ preparation across multiple exam boards, with question versioning, exam templates & bundles, difficulty tuning, timed assignments, subscription billing, and AI-assisted grading with speech input for question capture.
Tutor-side companion platform for authoring questions, building exam templates, running exam sessions, and managing student groups. Bulk PDF-to-question import with image processing and role-based auth.
Cross-platform mobile marketplace with a shop (cart, coupons, subscriptions), a six-step provider-booking flow across vets, grooming, boarding, walking, day care and training, and central pet profiles including health, ID and insurance.
Cross-platform pet social app with profile discovery, matching, chat, and an admin dashboard. Flutter client talking to a Next.js API backed by Supabase Postgres, region-pinned for ~250ms responses.
Cross-platform pet-care app with AI-driven care recommendations and a React / Supabase reference implementation for the web surface. Shipped as a Flutter mobile product with clean feature-first architecture.
Resume-matching product with LLM-based scoring, native mobile clients, and a Python back end. Bulk resume parsing, ranked shortlists, and recruiter workflows.
Production-shaped Python RAG pipeline built as clean, replaceable modules — PDF parser, chunker, embedder, vector store, and QA layer — so any stage can be swapped for a different model or vector index without rewriting the rest.
Scraping and analysis platform turning 14+ public sources into a queryable market and startup intelligence feed. Scheduled crawlers, dedupe, topic extraction, and structured search on top.
One product, three surfaces — a Python FastAPI back end shared by a Flutter mobile app and a native Android (Kotlin) client. Clean CRUD, auth, and models designed to scale to additional surfaces.
Multi-modal classification system fusing hyperspectral imaging, RGB computer vision, and depth into a single decision. Handles crumpled, overlapped, and mixed-material inputs that typically break conventional inline sorting — with IoT control loops for production-line deployment.
Stereo-camera ego-motion estimation (pitch and roll rate), optical-flow vector generation, image harmonisation across near-range multi-camera imagers, and image enhancement tuned for automotive latency and dashboard-display constraints.
Deep-learning pipelines for medical image classification, enhancement, and anatomical segmentation. Includes an automated testing protocol that measures the quantitative impact of upstream imaging on downstream CV processing — production-graded from research through validation.
Indoor visual-odometry pipeline fusing LiDAR, RGB, and IMU inputs. 2D-to-3D reconstruction and 3D-object segmentation using YOLO 7, NeRF, and EfficientNet, backed by octree-optimised point-cloud processing that runs in real time.
If your product looks like one of these — or something we haven’t shipped yet — let’s talk.
Start a project