The Forge

What Udjan does We build hydrogen augmentation devices for diesel engines. Installed on industrial generators, the device cuts fuel consumption and lowers emissions. The hardware is in the field. The clients are real. What we need now is the software layer that proves the savings with live data. What FuelSense must deliver The platform receives data from sensors installed on industrial diesel generators and turns it into proof. Measurements coming in: – Fuel flow, supply line. Liters per minute, continuous. – Fuel flow, return line. Liters per minute, continuous. Diesel engines send unburned fuel back to the tank. Actual consumption is supply minus return. – Power output. kW and kWh, three-phase, continuous. – Ambient temperature. For diesel density correction. – Engine coolant temperature. For efficiency drift detection. – Timestamp on every packet. UTC, second-level resolution. The environment Sensors connect to a site gateway over Modbus RTU and pulse input. The gateway connects to the internet over site WiFi. No cellular, no ethernet assumed. WiFi drops are expected. The gateway buffers locally and resyncs on reconnect. The platform handles out-of-order data without breaking the timeline. One gateway per site. Multiple sites per client. Data arrives as MQTT or HTTPS POST. The builder team chooses the protocol. The metric the platform computes Specific Fuel Consumption (SFC) = net fuel consumed (supply minus return) divided by kWh generated. Continuous. Time-series. Binned by load percentage, because SFC at 40 percent load and SFC at 80 percent load are not the same number. Before-Udjan vs after-Udjan curves at matched loads. This is the proof. What the platform must show Client portal (their generators only): – Live SFC, live kW, live fuel rate – Before vs after SFC curves at matched load bins – Savings in liters, rupees, CO2 – Historical trends. Daily, weekly, monthly. – Alerts when SFC drifts beyond threshold – AI insights tab. Natural language questions about their generators. Real answers from their data. Udjan team portal (fleet view): – All installations, all clients – Health of every gateway. Online, last seen, data gaps. – Fleet-wide savings aggregation – Per-site drill-down The build Backend plus cloud time-series store. Web dashboard, responsive. Mobile app for iOS and Android. Alerts over email and push. What Udjan provides – First site is live and waiting. Sensors installed, gateway on WiFi, data streaming. – Cloud infra, domain, hosting. Covered. – Second and third sites ready to onboard the moment the platform accepts them. What Udjan does not provide – Product recommendations on sensor or gateway hardware. That is not where this dare lives. – A written architecture. The builder team designs the platform. Who we are looking for A team of 2 to 3. Strong on backend and React / React Native. IoT or real-time data experience is a bonus, not a requirement. What you get – A real product with a real client. Not a hackathon project. – All infrastructure costs covered by Udjan. – Builder credit on the platform. – If FuelSense scales, an equity conversation. We mean it. How to respond Submit a proposal here on the Forge, or email support@udjan.com with subject "FuelSense Build Dare". Tell us who you are, what you have shipped, and how you would approach this. We will get on a call.

At Fidus Central, my office in Chandigarh, there is a room called Paunahari.Fidus. Inside it, an akhand jyot has been burning since March 2025 -- lit during the last Navratras. It has not gone out once. It cannot go out. I am a mechatronics engineer. I am never in one place -- multiple ventures, multiple cities, active training schedule. There is no watchman. The variables are real: wick condition, ghee level, weather, cold winter nights, long travel days. Build a system that monitors the health of this flame and alerts me before it is too late. WHAT NEEDS TO BE BUILT (V1): - Detects flame health -- not just presence or absence - Monitors intensity, stability, and ghee level indicators - Sends a phone alert before the flame is in danger - Runs autonomously. No manual check required. No human in the loop. WHAT A WINNING SUBMISSION LOOKS LIKE: - Working hardware installed and running at Fidus Central, Chandigarh - Alert reaches my phone reliably -- tested over minimum 72 hours - False positive rate is low enough to trust - Documentation of the build: components, wiring, code, setup instructions - Build cost within Rs. 3,500 (hardware covered by Fidus Synergies) This is not a simulation. Not a prototype on a desk. A real flame. A real room. A real dare.