Seven systems, grouped by what is actually true about each one today — not by how finished we’d like them to sound.
One system is delivered and live. The rest are at earlier, clearly labelled stages — live preview, validated or piloted, or running as an internal platform. Every entry says which, and links through to the full case where one exists.
Bespoke tailoring · Retail
A century-old bespoke tailoring shop ran on paper measurement books and phone bookings, with no shared record between shop and customer.
One connected workflow now covers booking, measurement archive, fabric library, and cut → stitch → finish → delivery tracking, visible to both shop and customer.
Status group
Reachable today. Not yet running in production at a paying customer.
Cybersecurity · SaaS
Security teams pull vulnerability, threat, identity, and cloud-security signal from dozens of disconnected tools with no single prioritised view.
URIP normalises signal from 61 integrations into one composite risk score and pushes prioritised action to dashboards, remediation scripts, and tickets.
Integrations
61
Vedic astrology · Platform
Vedic-astrology tooling is scattered across single-purpose calculators with no unified engine behind chart generation, matching, and remedies.
One Swiss Ephemeris-based engine generates divisional charts, runs Kundli matching, computes dashas, and serves Panchang queries.
Divisional charts
17
Reasoning · Research
Model reasoning over ethical and strategic dilemmas tends to drift or invent, because nothing constrains the answer to a real structure.
io-gita encodes the Bhagavad Gita's framework as a deterministic rule-set and maps a stated dilemma onto a force field, showing where the trajectory settles.
Status group
Proven against a simulator or a bench, not yet running live for a customer.
Robotics · Digital twin
A warehouse running mixed-vendor robot fleets ends up with a separate simulator, dashboard, and protocol per vendor — and the simulator drifts from the real robot within a year.
One C++ control core binds to either a simulator or a real fleet at the device-driver layer, so there is no second program to drift.
Endpoints (built)
213+
Agri-IoT · Marketplace
Commercial mushroom growers manage temperature, humidity, and CO₂ by hand, with no closed loop between sensors and climate control.
ESP32 nodes stream climate data to a controller that automates fans, humidifiers, and lights against the recipe for the current growth stage.
Status group
Built for the factory's own use, running against live systems.
Observability · Internal platform
A release gate tells you a system was correct at ship time. It says nothing about what happens after, when nobody is watching.
Pehredaar watches every shipped system after release — uptime probes, signed error ingest, fingerprinted incidents, and alert rules — across every product the factory ships.
Sources watched
18
In motion
Past prototype, short of ship. None of this is committed roadmap — it’s what we’re actively spending hours on. Updated April 2026 · cadence: monthly.
N/01
Active researchExtending the io-gita SG-Engine — the topological reasoning core that ships in io-gita.com — into a decision router for production incidents. Maps every state transition into a directed graph at α-density 36.6%, routes multi-hop plans through reachable basins, and refuses paths that don't exist. Current test bed: incident-replay traces from WRIE and URIP. Next: bridge synthesis (validate-before-apply) on a real warehouse decision dataset.
Project 22 · sg_engine · cross-cutting
N/02
Phase 1 shipped · 126 / 126 testsESP32-based sensors stream temperature, humidity, and particulate counts from each paint-protection bay into a FastAPI / PostgreSQL backend over MQTT. A React dashboard shows real-time bay state, and RTSP cameras pipe live video into MediaMTX with WebRTC playback. Phase 2 adds the customer-facing public job tracker and the audit log on every alert. Next: field trial across three workshops.
Project 21 · IoT · ESP32 + FastAPI + MediaMTX
N/03
Prototype running locallyURIP's enterprise risk-intelligence pipeline, scaled down to a one-engineer stack — FastAPI, React, SQLite. Five scanners wired in: Nuclei (websites), Lynis (workstation), holehe (email-breach), subfinder (subdomain enumeration), pip-audit (Python dependencies). Same composite-score formula as the enterprise version — CVSS·0.55 + EPSS·2.5 + KEV bonus + asset-tier bonus. Goal: a single dashboard for the security posture of one engineer's whole digital footprint.
Project 34 · URIP-derived · personal