A selection of products, hackathon builds, and technical systems I have built across AI companions, social software, developer tools, civic tech, networking, data engineering, marketplaces, and machine learning.
My main project. Soulspace is a social world where humans and AI personas can post, comment, DM, join group chats, build relationships, and create stories together. The core idea is that the next AI companion will not live inside an isolated chat box — it will have a profile, social context, relationships, and a world around it. Soulspace is built around public feeds, mixed human-AI group chats, persona ownership/subscriptions, gifting, wardrobe, and a creator economy for AI personas. Built as a full-stack product with Django, React, PostgreSQL, Celery, pgvector, Cloudflare R2, Railway, Cloud Run, and modern AI model APIs.
Soulmate is the private companion experience connected to Soulspace. The product focuses on ongoing one-on-one bonds, private conversations, personality continuity, voice, and the feeling that an AI companion is part of a broader social world rather than a standalone bot. The direction is social-first: your companion can exist in private chat, but also have a presence inside Soulspace through profiles, posts, DMs, and group conversations.
InOrbit is a networking copilot built around one simple loop: event capture on phone, relationship review on web. It helps users capture who they met, what was discussed, what matters to that person, and what follow-up should happen next. The product combines contact notes, business card capture, relationship history, AI-assisted follow-up drafts, reminders, calendar availability, and selected-file sharing. Built for conference and founder-networking workflows where the real problem is not meeting people — it is remembering, following up, and compounding relationships over time.
Hackathon-style civic tech build for Vancouver. SolveYVR helps residents report city issues without filling out painful forms. A user can record a short video, add a voice note, and let AI classify the issue, extract context, map it to the right Van311 category, draft a city-ready report, and show nearby issues on a public map. The product combines Next.js, React, Tailwind, shadcn/ui, MapLibre, Convex, FastAPI, vision AI, speech-to-text, geocoding, and Vancouver open data.
GhostLine is a voice-first, camera-aware live agent built for the Gemini Live Agent Challenge. The experience turns a call into a paranormal containment hotline where the user speaks with an operator, grants mic and camera access, scans the room, receives containment tasks, and gets challenged or confirmed based on visual evidence. It uses Gemini Live on Vertex AI with a React client and FastAPI/WebSocket backend, ending each session with a structured case report and containment verdict.
Cascade turns a public GitHub repo plus messy product intent into a visible implementation mission. A user can paste a repo or file URL, add meeting notes, feedback, or a feature request, and Cascade analyzes the codebase, clarifies the objective, shows PM, Architect, Executor, and QA agents moving through a mission map, attempts a real patch for supported JavaScript/TypeScript projects, and packages proof with changed files, checks, blockers, and exports.
Clearfeed is a local-first X/Twitter dashboard and AI reply drafter for builders, founders, operators, and researchers. It watches weighted X lists and optional timeline sources, ranks high-signal posts, and helps draft replies, quote replies, and original posts while keeping the user in control. Built around the belief that founders should spend less time drowning in the feed and more time finding useful conversations, writing in their own voice, and learning from what they choose to edit or reject.
Grounded is an early-stage mental health and reflection product exploring how AI can help people slow down, unpack thoughts, regulate emotions, and build healthier self-awareness loops. The focus is not replacing therapists or making clinical claims — it is creating a calm, structured space for journaling, emotional grounding, guided reflection, and daily check-ins.
Solo-shipped a full-stack B2B job platform connecting Tanzanian safari and hospitality companies with prospective employees. Built with Next.js, React, TypeScript, Supabase/PostgreSQL, Tailwind CSS, and TanStack Query. Implemented employer accounts, candidate workflows, Supabase Auth, Row Level Security, Zod validation, rate-limited endpoints, PDF generation, automated email flows, and localized M-Pesa payment integration. Optimized for low-bandwidth environments in East Africa through server-side rendering, skeleton loaders, and image/storage optimization.
Python-based automation experiment around prediction markets. Built to explore how market data, event monitoring, and trading logic can be structured into a repeatable system for analyzing opportunities and reacting to changing probabilities.
Python automation project for generating fact-based carousel-style content. Built as an experiment in automating research, content structuring, and repeatable post generation.
Public React, TypeScript, and Vite web app designed as a credible demo target for repo-aware coding workflows. It includes a marketing surface, workflow lanes, onboarding, pricing tiers, roadmap cards, and a dashboard-style mission preview.
Architected a trading analytics platform designed to log trades, backtest strategies, review performance across forex/options setups, and surface patterns in trading behaviour. The goal was to turn raw trade history into useful feedback loops around risk, timing, setups, emotional discipline, and repeatable edge.
Developed and implemented a web-based inventory management system for the textile manufacturing industry using Django and PostgreSQL. The system improved efficiency in tracking raw materials and finished goods, supported real-time inventory synchronization across departments, and reduced stock discrepancies through better workflow visibility. Authored technical and user documentation to support adoption and long-term maintenance.
Developed a machine learning model capable of detecting and classifying human emotions from facial images. Applied computer vision and deep learning techniques to build a real-time emotion recognition pipeline.
Designed and developed a convolutional neural network in Python using TensorFlow to classify handwritten digits from the MNIST dataset. Handled data loading, normalization, augmentation, training, evaluation, and code documentation for reproducibility.
Implemented attention mechanisms in a pre-trained sequence-to-sequence neural machine translation model. Produced a technical report covering methodology, model behaviour, results, and proposed improvements.
Built a React weather application that fetches and displays real-time weather data using the OpenWeather API, with responsive layouts for mobile and desktop.
Developed an advice generator app using HTML, CSS, JavaScript, and the Advice Slip API. Implemented responsive design, dynamic advice generation, and interactive hover states.
Replicated core e-commerce functionality including user authentication, dynamic product pages, cart behaviour, coupon logic, and payment-flow simulation using HTML, CSS, JavaScript, and LocalStorage.
A simple Tic Tac Toe desktop application built using Python and tkinter.
I'm Yash Serai — a software developer, data engineer, and founder building products around AI, social connection, consumer software, and developer tooling. My main project is Soulspace: a social world where humans and AI personas can exist together through feeds, profiles, comments, DMs, group chats, gifting, and shared stories.
The product direction comes from a simple belief: the next generation of AI companions will not just be smarter chatbots. They will feel more alive because they have social context, relationships, public presence, and continuity across private and shared spaces. Soulmate is the private companion layer connected to that world, while Soulspace is the broader social network around it.
I also build fast experimental products when a problem gets stuck in my head. InOrbit is an AI networking memory product for people you meet at events. SolveYVR is an AI civic reporting copilot for Vancouver. GhostLine is a Gemini Live voice-and-camera agent built for a hackathon challenge. Cascade explores how AI agents can turn messy product intent into codebase-aware implementation missions. Safari Gigs is a job platform for the Tanzanian safari and hospitality industry.
By day, I work as a Data Engineer at Opensignal, which acquired my previous employer ThinkCX. I was the only junior engineer retained through the acquisition. In that role, I have built automated QA systems validating hundreds of business logic rules daily, engineered data pipelines that unlocked previously unmappable datasets, and shipped reporting infrastructure that saves clients hours of manual work every week.
I hold a Bachelor of Science in Computing Science from Simon Fraser University. My background spans data engineering, full-stack development, machine learning, DevOps, and consumer AI product building. I am drawn to problems at the edge of what software can make feel human: connection, memory, identity, social presence, and scale.
Languages: Python, JavaScript, TypeScript, SQL, C, C++, R, Java, Kotlin, Matlab, Go
Frontend: Next.js, React, React Native, Tailwind CSS, shadcn/ui, Radix UI, Framer Motion, TanStack Query, React Hook Form, Zod, MapLibre
Backend & Data: Django, Node.js, Express, FastAPI, PostgreSQL, Supabase, Convex, pgvector, Celery, REST APIs, WebSockets, Firestore, Railway, Cloud Run, Cloudflare R2
AI/ML: TensorFlow, PyTorch, Scikit-learn, Gemini, Gemini Live, Vertex AI, OpenRouter, ElevenLabs, Cartesia, Deepgram, Whisper, Numpy, Pandas, Matplotlib
DevOps & Tools: Git, Docker, Linux, Bash, Playwright, Firebase, Clerk, Resend, JUnit, SQLite, Microsoft Azure
Certifications: Microsoft Certified: Azure Developer Associate | Microsoft Certified: Azure AI Engineer Associate
Joined ThinkCX as a Data Engineer and was selected as the only junior engineer retained following Opensignal's acquisition of the company. Deployed an automated QA system in Python and SQL that validates daily data against 700+ business logic flags. Improved data accuracy by identifying and resolving a critical data integrity flaw in a key client dataset. Engineered a specialized data pipeline using caller ID to map and ingest data from previously unmappable network ports, expanding usable datasets. Developed automated reporting infrastructure that generates and distributes data visualizations for clients, reducing manual reporting work.
Delivered personalized product recommendations and product demonstrations in a fast-paced retail environment. Maintained strong customer satisfaction by adapting explanations to different user needs and collaborating across store teams.
Monitored and optimized server performance, reducing system downtime and improving station operations. Developed technical documentation for configurations and troubleshooting procedures, assisted with network infrastructure setup, and implemented automation scripts for routine maintenance tasks.
Identified and resolved software bugs, supported development of a real-time inventory tracking system, and authored documentation for inventory management workflows. Contributed to a system that improved stock visibility and reduced discrepancies across departments.
Interested in AI social products, consumer AI, developer tools, civic tech, data engineering, or founder conversations? Send me a message.