Projects

EquiSmart

Running today and in regular use; it is not an open public service you can sign up for.

Paper-first algorithmic trading for Indian markets, with a live screener and broker connect.

Retail traders lack an affordable, paper-first way to test and run algorithmic strategies on Indian markets.

A multi-tenant SaaS that connects to a broker for paper-first trading strategies, runs an NSE stock screener, and shows a live trading-terminal chart.

• A scheduled strategy engine ticks every 15 seconds.
• Live order placement needs an explicit opt-in flag and is off by default, so the platform's default behaviour is strategy testing, not automated live-money trading.

• FastAPI
• Postgres
• Celery/Redis
• Next.js

AutoMotivate workshop

The core works end to end and can be demonstrated; it is not production-hardened or generally available.

A delivery-ready two-day GenAI workshop build for automotive R&D teams, and the one carded build behind AutoMotivate, EmbedAI's AI upskilling practice for automotive R&D.

Automotive engineering teams are told to adopt AI, then handed a generic course built for office workflows, not their own kind of work.

An 11-module, 22-live-demo executive workshop with a full delivery kit (run-of-show, preflight checklist, governance pack) covering GenAI-assisted coding, ECU log analysis, agentic workflows and PII-safe automation, all on synthetic automotive data for a fictional company.

• Built on LangGraph and CrewAI for the agentic-workflow modules, with a PowerShell workshop runner orchestrating the demo scripts.
• 15 or more synthetic automotive artefacts are all authored for a fictional company, so the demo content itself is safe to describe in full.

• PowerShell workshop runner
• Python demo scripts
• LangGraph
• CrewAI

Cockpit

Running today and in regular use; it is not an open public service you can sign up for.

A task-dispatch and status dashboard across local project repositories.

Coordinating work across many local project repositories and tracking AI-assisted development hours otherwise needs manual bookkeeping.

Assigns tasks to local project repositories, drains them into each repository's own task list, and reports project status and AI coding-session hours from one dashboard.

• The dispatch loop is built as a request/response contract across projects (assign, pick, fan-out, release), designed to run without a central always-on service holding state.

• Python (standard library) behind a single-page HTML/JS UI
• scheduled background jobs

No public artefact yet.

Capabilities

Running today and in regular use; it is not an open public service you can sign up for.

Two client-proposal demos for manufacturing clients: a design-for-X (DFX) knowledge platform, and an AOI-window defect classifier.

Manufacturing and EMS clients need to see and pilot practical AI — design-rule capture, defect classification — before committing to a larger engagement; a slide deck does not answer that, a working demo does.

The DFX platform runs a design-rule engine and a Q&A feature that answers only from cited sources (it visibly declines to answer outside that scope). The AOI-window classifier is a CPU-based defect classifier that sorts inspection-window images into defect classes with a confidence score. Both are paired with proposal decks generated from code.

• The Q&A feature's citation-or-refusal behaviour is the notable engineering point: it is built to say "outside my sources" rather than guess, backed by 10 passing tests.
• The classifier runs on CPU (ONNX/torch), not dedicated inference hardware, which matters for a client who does not want to add a GPU to a line, backed by 65 passing tests.

• FastAPI backend
• SQLite full-text search
• a CPU-based ONNX/torch classifier
• Node.js generated proposal decks

Content Creator

Running today and in regular use; it is not an open public service you can sign up for.

Turns one idea into publish-ready content packages for ten platforms, with a human approval gate.

Producing consistent, on-brand content across many social platforms from a single idea is slow and error-prone.

A git-versioned pipeline that carries an idea through qualification, scripting and drafting, then packages it per platform (YouTube, Shorts, Instagram, TikTok, X, LinkedIn and more), gated by human review before anything ships.

• Ten platform adapters, each with its own spec and prompt.
• The human-approval gate sits before anything ships, not after, so nothing is published automatically.

• Python
• Claude Code skills and a platform-writer subagent
• optional local media tooling

DayScribe

Running today and in regular use; it is not an open public service you can sign up for.

An offline-first life-logging app that tracks all 24 hours of the day against a personal values tree, in a hosted-ready edition and a local-first edition.

People who want to understand how their time actually aligns with what they value lack a low-friction, always-available way to log it.

Logs whole-day time against a category tree, tracks habits and counts, and keeps notes, tasks and a journal. It syncs across a person's own devices with full offline support. The local-first edition does the same over the local network, with no hosted backend required.

• The two editions are a real engineering claim about offline-first design, not a restatement: one runs hosted behind an API, the other runs local-only with network sync, from the same taxonomy and interface.

• vanilla JavaScript PWA (hosted edition) with a Python API and Postgres
• vanilla JavaScript PWA (local-first edition) with a lightweight local sync gateway

No public artefact yet.

• DayScribe — local-first edition In use

DashCam

Running today and in regular use; it is not an open public service you can sign up for.

A pipeline that ranks, transcribes, speaker-labels and mood-tags dashcam footage, so a moment can be found by the words spoken. A companion Android app does the same recall entirely on the phone. No clip, frame or still is shown here: the project's own policy blocks publishing footage until face and plate blurring and speaker consent are in place.

FilmyGyan

The core works end to end and can be demonstrated; it is not production-hardened or generally available.

Turns Indian-cinema dialogue into a searchable, always-attributed knowledge base and persona chat.

Finding "the scene where a character says the thing that fits this situation" is otherwise a manual, memory-dependent search through films.

Ingests film captions into a tagged dialogue knowledge base, powers situation-based search, and runs a persona chat that always attributes its quotes back to the source film.

• The always-attribute-the-source behaviour is the point, given the dialogue is third-party copyrighted material.
• The production tier caps quote length rather than storing full text, so the site describes the search and attribution capability and reproduces no quoted dialogue.

• Python
• SQLite
• TF-IDF retrieval
• optional AI-assisted tagging

From TT Global Studio

Built far enough to prove the idea, then paused; it is not in active development.

A dormant worktree of KinoAI. It finds the moments in coaching video where a coach demonstrates a stroke, then aggregates a per-stroke, joint-angle body-position template from them. It is mentioned beside KinoAI because it shares that knowledge base and taxonomy, not as a separate product.

Guruji

Designed and specified in writing; no working software exists yet.

A plan for a personal AI tutor that keeps pace with what a student's own school is actually teaching that week. It combines a daily practice loop, spaced repetition and a Socratic tutor chat that never gives away answers on assessed items. It is a design only, with no working software behind it.

WoodTech Interior Design

The core works end to end and can be demonstrated; it is not production-hardened or generally available.

Turns a furniture idea into an exact cut-list, bill of materials and cost, with live 3D and AR.

Going from "I want a wardrobe like this" to an exact, buildable manufacturing spec and price is normally a slow back-and-forth with a carpenter.

Lets a user configure a wardrobe, TV unit or kitchen, then generates an exact panel cut-list, bill of materials, cost estimate and a live 3D view. It can turn a floor plan into whole-house cabinetry. A chat assistant and an Android AR app place the design at true scale.

• Automated cut-list tests pass against golden files within 0.5 mm tolerance — the kind of accuracy claim a manufacturing-minded peer checks first.

• a Python API
• a React and Three.js web app
• a Kotlin/Compose Android AR app

KinoAI

The core works end to end and can be demonstrated; it is not production-hardened or generally available.

An index of exactly which coaching video teaches a given table-tennis skill, built toward a camera-only AI coach.

There is a huge amount of table-tennis coaching video online, but no way to find the exact clip that teaches a specific skill or fault fix.

Ingests public coaching-video transcripts, tags them against a stroke/skill taxonomy, and builds a searchable index of teaching moments — the knowledge layer for a planned camera-only AI coaching product.

• A 239-node taxonomy with 1,797 lexicon terms tags 39,779 cues, covering 366 of 459 known videos.
• The camera-only coaching product itself remains a plan, not a build, and the copy says so.

• a Python transcript pipeline
• a stroke/skill taxonomy
• scheduled ingestion

Personal Assistant

Running today and in regular use; it is not an open public service you can sign up for.

A personal capture-to-knowledge pipeline, plus the reading layer that makes the result searchable.

Notes, ideas and progress captured throughout the day get lost between different apps and never make it into a durable, organised knowledge base.

Captures notes and files from multiple sources into one inbox, automatically drains them into a daily journal, and promotes reviewed material into a knowledge base. The reading layer then renders that knowledge base as a browsable, installable site with offline full-text search, backlinks and a link graph.

• The reading layer builds the whole knowledge base into a static site with its own search index and link graph, backed by 44 automated checks. That offline index and the installable PWA around it are its own engineering, not a wrapper over a hosted search service.

• Python (standard library) with scheduled automation and a knowledge-base vault for the capture side
• Astro static-site generation with an offline search index for the reading layer

• Personal Assistant — reading layer In use

SBRP

Designed and specified in writing; no working software exists yet.

A design for a WhatsApp/Telegram-driven operations app for interior-hardware shops. Staff log work by voice or text, orders move through a tracked lifecycle, and the system estimates the materials each job needs. Nothing is built yet.

Triage

Designed and specified in writing; no working software exists yet.

A planned digest that turns a personal inbox and phone notifications into short, privacy-scrubbed summaries. Sensitive details such as one-time codes and card numbers are removed before a digest is produced. Every module is a stub today, so nothing runs yet.

xPilot

The core works end to end and can be demonstrated; it is not production-hardened or generally available.

A voice-first driving copilot for Indian roads that keeps working without a network connection.

Existing voice assistants for driving assume constant connectivity and don't handle India-specific driving needs well.

A tap-to-talk Android copilot that answers questions over a live map during a trip, using on-device rules and a local AI model first and falling back to a cloud AI model only when needed.

• The local-first, cloud-fallback design is the point: it answers from on-device rules and a local model before ever calling out, so it degrades gracefully rather than failing outright when the network drops.
• The cloud fallback carries a free-tier daily request limit, so the copilot is not positioned as unlimited cloud-dependent capability.

• a Kotlin/Compose Android app running an on-device language model
• a cloud AI model as a fallback