Sabyasachi Panigrahi

Founder • Engineer • Product Builder • IoT & AI Systems Architect

More than a decade of experience building real-world systems—from software products and consumer platforms to autonomous IoT infrastructures and AI-powered public-safety solutions. My work designs systems that operate under real constraints: unreliable networks, harsh environments, human unpredictability, and safety-critical conditions.

"If the internet fails, sensors drift, humans forget, or conditions change the system must still behave safely and intelligently."

Companies Built

Building companies from the ground up—across technology, product, operations, and day-to-day decision making.

Prixso Software

Founder · Engineering & Systems Foundation

Prixso Software was a full-stack software engineering company that became the proving ground for building real, production-grade systems. The work ranged from web platforms and mobile applications to early IoT and automation projects, all built under real client constraints such as timelines, budgets, and long-term maintainability.

Node.js Python React PostgreSQL Docker AWS ESP32
API Architecture & Integration
  • RESTful API design with authentication, authorization, and rate limiting
  • Real-time WebSocket connections for live system updates
  • Third-party integrations including payments, messaging, and cloud services
  • Service-oriented and microservice-style architectures where required

What We Did

  • Built custom web applications with complex business logic
  • Developed backend platforms and APIs supporting real-world usage patterns
  • Created mobile applications with offline-first and sync-based designs
  • Integrated legacy systems with modern software stacks
  • Delivered early-stage IoT and automation solutions for connected devices

My Role & Responsibility

As the founder, my role extended well beyond engineering. I handled early client discussions, requirement discovery, proposal preparation, pricing conversations, and delivery commitments. On the technical side, I designed system architectures, wrote and reviewed core code, defined engineering standards, and stayed involved through deployment and maintenance. At the same time, I was responsible for balancing timelines, managing trade-offs, and ensuring the company could sustainably deliver reliable systems.

Key Learnings

  • How real systems fail under load and imperfect conditions
  • Why clean architecture and thoughtful abstraction matter long-term
  • How to translate unclear business needs into dependable software
  • The importance of monitoring, logging, and operational visibility
  • Designing databases and APIs for growth rather than short-term fixes

Tifinco Foodtech

Founder · Product Thinking & Subscription Economics

Tifinco was a meal subscription platform built to address the chaos and decision fatigue of on-demand food ordering. Instead of focusing on one-off orders, the platform emphasized planned meals, predictable operations, and long-term customer retention through subscription-based delivery.

React Native Node.js MongoDB Payment Gateways Firebase Google Maps API
API & Integration Layer
  • Subscription management APIs with automated billing cycles
  • Payment gateway integrations with webhook-based reconciliation
  • Notification systems for order updates and reminders
  • Delivery routing and ETA calculation using location services
  • Analytics pipelines for understanding usage and churn patterns

What Made Us Different

  • Weekly and monthly meal subscriptions instead of daily ordering decisions
  • Controlled kitchen operations rather than restaurant aggregation
  • Customizable meal plans based on dietary needs and preferences
  • Predictable delivery windows that reduced customer anxiety
  • Clear portioning and transparent meal information

My Role & Responsibility

As the founder, I was involved across product, operations, and execution. I worked on pricing models, subscription structures, and customer experience, while also spending time understanding kitchen workflows, delivery constraints, and unit economics. On the product side, I designed UX flows, prioritized features, and built systems that supported subscriptions, customization, and retention. Many decisions were shaped by on-ground realities—what could actually work in daily operations, not just what looked good conceptually.

Key Learnings

  • Small UX friction directly impacts conversion and retention
  • Pricing perception is as important as pricing itself
  • Operations and software must evolve together
  • Retention matters more than acquisition at scale
  • Customer support feedback is a critical product signal
  • Trust and transparency are essential in subscription businesses

Home Automation

Context-aware systems that understand presence, environment, and act autonomously while respecting human overrides.

Smart Home Automation Platform

Context-Aware, Presence-Driven, API-First Architecture

A complete home automation system that goes beyond simple timers and switches. This platform uses multi-sensor fusion to understand room occupancy, ambient conditions, and user patterns to make intelligent decisions about lighting, ventilation, and climate control—all while functioning perfectly offline.

ESP32 MQTT Node-RED InfluxDB Grafana Home Assistant
API Ecosystem
  • RESTful Device Control API for third-party integration
  • MQTT broker for real-time device communication
  • WebSocket API for live status updates on dashboard
  • Scheduling API with cron-like expressions for automation rules
  • Webhook support for external triggers (weather, calendar events)

Problem Statement

Most commercial smart homes are cloud-dependent, unreliable when internet drops, and rely on primitive timer-based automations. They can't distinguish between someone actually being in a room versus a pet wandering through, leading to energy waste and user frustration.

Hardware Architecture

  • ESP32/ESP8266 microcontroller nodes deployed in each room with local processing
  • mmWave radar sensors for precise human presence detection (detects breathing, even when stationary)
  • PIR motion sensors as secondary validation layer to prevent false positives
  • BH1750 ambient light sensors for adaptive lighting control
  • DHT22 temperature and humidity sensors for climate awareness
  • Physical toggle switches wired as hard overrides (always respected by automation)
  • Relay modules for AC appliance control with electrical isolation

🏗️ System Architecture: Edge Processing → MQTT Mesh Network → Central Hub → Cloud Sync (Optional)

Edge Intelligence & Decision Logic

Decision Engine: Lighting/Climate State = f(Presence Confidence, Ambient Light Level, Temperature, Humidity, Time of Day, Manual Override State)

The system uses multi-sensor confirmation with weighted confidence scoring. For example, mmWave detection (70% confidence) + PIR trigger within 5 seconds (20% confidence) + light level below threshold (10% confidence) = Action. Time-decay logic ensures rooms don't stay lit if presence signals stop.

Zero energy waste from forgotten devices No user micromanagement needed Natural, invisible automation Privacy-first (no cameras)

Intelligent Bathroom Automation

Safety-Critical, Multi-Sensor Fusion System

A comprehensive bathroom automation system designed with safety as the top priority. This system manages ventilation, lighting, and environmental monitoring while actively protecting against gas leaks, electrical hazards, and humidity-related issues like mold growth.

ESP32 MQ-2 Gas Sensor DHT22 Relay Control MQTT Watchdog Timer
Safety & Control APIs
  • Emergency alert API with SMS/push notification webhooks
  • Sensor health monitoring API with automatic diagnostics
  • Manual override API with time-limited control
  • Historical data API for humidity/temperature trend analysis
  • System status API exposing all sensor readings in real-time

Risk Factors Addressed

  • Gas leakage from water heaters or LPG lines (potentially fatal)
  • Electrical hazards from moisture exposure in enclosed spaces
  • High humidity leading to mold growth and respiratory issues
  • Poor ventilation causing CO₂ buildup and discomfort
  • Temperature extremes during shower usage

Safety Engineering Principles

  • Fail-ON Logic for Critical Systems: If gas sensor detects leakage OR sensor communication fails, exhaust fan IMMEDIATELY turns ON and stays ON until manually cleared. Safety failures always bias toward action, not inaction.
  • Sensor Sanity Checks: Every sensor reading is validated against physically possible ranges. A humidity reading of 110% or -20°C indicates sensor failure, triggering a fail-safe mode and user alert.
  • Watchdog-Based System Reset: If the main control loop hangs or crashes, a hardware watchdog timer automatically resets the system to a known-safe state (all ventilation ON, alerts sent).
  • Manual Override with Auto-Revert: Users can manually control devices, but the system automatically resumes automation after 30 minutes or when the user leaves, preventing forgotten overrides.
  • Redundant Communication: Critical alerts are sent via multiple channels (local siren, mobile push, SMS) to ensure notification delivery even if one path fails.
Improved air quality and comfort Significant energy efficiency gains Proactive safety protection Zero maintenance intervention needed

FarmTech

Autonomous farming systems with closed-loop climate intelligence and real-time monitoring.

Autonomous Mushroom Farming System

Closed-Loop Climate Intelligence with PID Control

A fully autonomous environmental control system for commercial mushroom cultivation. Mushrooms are extremely sensitive to CO₂ levels, humidity, and temperature—even small deviations can destroy entire batches. This system maintains optimal growing conditions 24/7 with minimal human intervention, dramatically improving yield consistency and reducing contamination risk.

ESP32 MH-Z19B CO₂ Sensor DHT22 Ultrasonic Misting PID Controller LoRaWAN
Data & Control APIs
  • Real-time environmental monitoring API with 30-second intervals
  • Actuator control API with rate-limiting to prevent equipment damage
  • Growth stage management API with preset profiles per mushroom species
  • Alert API for contamination risk indicators
  • Historical analytics API for yield correlation analysis
  • Remote dashboard API for farmer mobile app integration

Environmental Parameters Monitored

  • CO₂ Concentration: Must stay between 800-1200 ppm during fruiting stage (too high = stunted growth, too low = wasted ventilation energy)
  • Relative Humidity: 85-95% for most species (lower = dried pins, higher = bacterial contamination)
  • Temperature: Species-specific ranges (oyster: 18-24°C, shiitake: 12-18°C)
  • Air Exchange Rate: Calculated from CO₂ decay rate to ensure fresh air circulation

Actuator Systems

  • Ultrasonic Misting System: Variable duty-cycle control based on humidity deficit. Prevents over-misting which causes bacterial blooms.
  • Ventilation Fans: Multi-speed exhaust and intake fans for controlled air exchange without shocking the mycelium.
  • Heating/Cooling: Dual-mode climate control with deadband zones to prevent oscillation and energy waste.
  • Air Circulation Fans: Low-speed continuous operation to prevent CO₂ stratification in corners.

Intelligence Layer: Adaptive PID Control

The system uses three cascading PID controllers for humidity, temperature, and CO₂. Each parameter has setpoint scheduling that adjusts based on growth stage (colonization, pinning, fruiting, harvesting). The controllers use historical performance data to auto-tune their gains, learning the specific thermal and humidity response characteristics of each growing room.

Consistent, predictable yields Minimal manual intervention Reduced contamination losses Faster cultivation cycles

Biofloc Fish Farming System

Real-Time Aquatic Life Support & Predictive Analytics

An advanced aquaculture monitoring and control system designed for high-density biofloc fish farming. In biofloc systems, water quality can deteriorate rapidly—dissolved oxygen crashes can kill an entire pond in hours. This system continuously monitors critical parameters and takes predictive action before conditions become life-threatening.

Arduino Mega DO Sensor (Atlas Scientific) pH Sensor Ammonia Sensor Turbidity Sensor 4G Connectivity
Water Quality APIs
  • Real-time water parameter API with sub-minute sampling
  • Predictive depletion API using trend analysis (alerts 2 hours before critical DO levels)
  • Feeding schedule API with adaptive recommendations based on water quality
  • Aeration control API with emergency override protocols
  • Historical correlation API linking water quality to fish health/mortality
  • Multi-pond dashboard API for farm-wide monitoring

Critical Parameters Monitored

  • Dissolved Oxygen (DO): Must stay above 4 mg/L (below 3 mg/L = fish stress, below 2 mg/L = mass mortality within hours)
  • pH Levels: 6.5-8.5 range (sudden changes indicate ammonia spikes or algae crashes)
  • Ammonia (NH₃): Toxic above 0.5 ppm—early warning of overfeeding or biofilter failure
  • Nitrate/Nitrite: Indicates nitrogen cycle health (nitrite spikes = incomplete nitrification)
  • Turbidity (NTU): Measures biofloc density—optimal range ensures bacterial protein production without oxygen depletion
  • Water Temperature: Affects DO solubility and fish metabolism (1°C change can alter feeding requirements by 10%)

Predictive Intelligence Model

The system doesn't just react—it predicts. By analyzing the rate of change of DO levels throughout the day, the system can forecast oxygen depletion 2-3 hours in advance. If DO is dropping at 0.3 mg/L per hour and current level is 4.5 mg/L, the system predicts critical levels in 5 hours and starts preemptive aeration before fish show stress symptoms.

Dramatic mortality reduction Lower feed wastage Healthier, faster-growing ponds Predictive crisis prevention

AgriTech

Data-driven precision agriculture and smart resource management systems.

Smart Agriculture & Precision Farming

Data-Driven Crop Intelligence & Resource Optimization

A comprehensive precision agriculture platform that transforms traditional farming into a data-driven science. By continuously monitoring soil conditions, weather patterns, and crop requirements, this system optimizes water and fertilizer usage while maximizing yield potential. Farmers make decisions based on real-time data, not guesswork.

ESP32 Capacitive Soil Moisture NPK Sensor Weather API Integration Solenoid Valves Solar Powered
Agriculture Intelligence APIs
  • Soil telemetry API providing NPK, moisture, pH, and temperature data
  • Weather integration API (OpenWeatherMap) for rainfall prediction and evapotranspiration calculation
  • Irrigation scheduling API using crop coefficients and soil water balance models
  • Fertigation control API with nutrient depletion tracking
  • Crop advisor API providing growth stage recommendations
  • Yield prediction API using historical data correlation

Core Sensing Infrastructure

  • Soil Moisture Sensors: Capacitive sensors at multiple depths (15cm, 30cm, 60cm) to track root zone hydration and drainage patterns
  • NPK Nutrient Sensors: Real-time nitrogen, phosphorus, and potassium level monitoring to guide fertilizer application timing and quantity
  • Soil pH Monitoring: Continuous pH tracking to optimize nutrient availability (most nutrients unavailable outside 6.0-7.0 range)
  • Ambient Sensors: Temperature, humidity, and solar radiation for calculating crop water requirements (evapotranspiration)
  • Weather API Integration: Connects to forecast services to prevent unnecessary irrigation before predicted rainfall

Intelligent Decision Logic

Irrigation Control: Water delivery = f(Current Soil Moisture, Root Zone Deficit, Evapotranspiration Rate, Weather Forecast, Crop Stage)

The system doesn't water on a schedule—it waters when crops need it. By combining real-time soil moisture with calculated water loss (based on temperature, humidity, wind, solar radiation), the system knows exactly how much water the crop has consumed and replaces only what's needed. If rain is predicted within 24 hours, irrigation is delayed automatically.

Fertigation Strategy: Nutrient Application = f(Current NPK Levels, Nutrient Depletion Trend, Crop Growth Stage, Yield Target)

Fertilizer is expensive and over-application pollutes groundwater. The system tracks nutrient depletion rates and applies fertilizer in small, frequent doses matched to crop uptake patterns. Fast-growing stages get more nitrogen, flowering stages get more phosphorus—all automatically adjusted.

Massive water conservation Optimized fertilizer efficiency Predictable, increased yields Reduced environmental impact

Multi-Tank Single-Pump Water System

Autonomous Distribution with Fair Energy Accounting

An innovative water distribution system serving multiple households with a single shared pump, but with intelligent priority management and transparent energy cost allocation. This system solves the common problem in shared water infrastructure: who pays for what, and how to ensure fair distribution without conflicts.

ESP8266 Ultrasonic Level Sensors Current Sensors (SCT-013) Motorized Ball Valves MQTT Broker Mobile Dashboard
Water Management APIs
  • Real-time tank level monitoring API for all connected tanks
  • Pump control API with priority queue management
  • Energy metering API attributing electricity costs per tank fill operation
  • Valve control API for automated water routing
  • Usage analytics API showing consumption patterns per household
  • Billing API generating fair cost breakdown based on actual energy usage

System Architecture

The system connects 4-8 overhead water tanks to a single borewell pump through electronically controlled valves. Each tank has an ultrasonic level sensor and dedicated valve. The pump has a current sensor connected to each tank's electric meter for precise energy attribution.

Intelligent Control Logic

  • Continuous Level Monitoring: All tanks report levels every 30 seconds. System tracks fill rates and consumption patterns.
  • Priority-Based Selection: When pump starts, system opens valve to the tank with lowest water level. If multiple tanks are critically low (<20%), system can open multiple valves simultaneously if pump capacity allows.
  • Dry-Run Detection: If pump current drops below threshold for 30 seconds, indicates no water in borewell. System immediately shuts down pump to prevent damage and sends alert.
  • No-Rise Timeout Protection: If tank level doesn't increase after 5 minutes of pumping, indicates valve failure or pipe blockage. System closes that valve and tries next tank in priority queue.
  • Overflow Prevention: When tank reaches 95% capacity, valve closes automatically. System includes 5% buffer to account for sensor uncertainty and valve response time.

Fair Energy Attribution System

This is the breakthrough feature. Each time a specific tank is filled, the system logs: (1) Pump runtime (seconds), (2) Average current drawn (amps), (3) Energy consumed (kWh = Voltage × Current × Time), (4) Timestamp and tank ID.

At month-end, the system generates a transparent report: "Tank A consumed 45 kWh, Tank B consumed 38 kWh, Tank C consumed 52 kWh" with exact timestamps. Electricity bill is divided proportionally. No more arguments about "who used more water"—data doesn't lie.

Fair energy cost distribution Zero overflow waste Fully autonomous operation Eliminates household conflicts

Industrial Automation

Utility-grade infrastructure, public safety systems, and remote healthcare solutions.

NB-IoT Smart Water Metering

Utility-Grade Infrastructure for Smart Cities

A next-generation water metering system designed for municipal utilities and large residential complexes. Using NB-IoT (Narrowband IoT) technology, these meters provide real-time consumption data, detect leaks and theft, and enable remote control—all while operating for years on battery power in underground deployment conditions.

NB-IoT Module (Quectel BC95) Ultrasonic Flow Sensor Motorized Valve LoRa (backup) AES-256 Encryption Li-SOCl₂ Battery
Utility Management APIs
  • Real-time consumption API with hourly usage data
  • Anomaly detection API identifying leaks, illegal usage, and meter tampering
  • Remote valve control API for shut-off during non-payment or emergencies
  • Billing integration API compatible with existing utility management systems
  • Network management API monitoring signal strength and device health
  • Bulk provisioning API for large-scale deployment management

Core Capabilities

  • Real-Time Consumption Tracking: Hourly usage data transmitted automatically, eliminating need for manual meter reading and enabling dynamic pricing models
  • Leak Detection: Continuous flow during night hours (2AM-5AM when usage should be minimal) triggers automatic leak alerts, potentially saving thousands of liters
  • Theft Detection: Sudden usage spikes or flow patterns inconsistent with billing history indicate illegal pump connections or meter bypass attempts
  • Remote Valve Control: Motorized ball valve can shut off water supply remotely for non-payment, maintenance, or emergencies—no technician dispatch needed
  • Predictive Maintenance: Flow sensor degradation and valve health monitored continuously; alerts sent before complete failure

Why NB-IoT for Water Meters?

NB-IoT is purpose-built for smart utility applications. It provides: (1) Deep indoor/underground penetration (20dB better than LTE), (2) Ultra-low power consumption (10+ year battery life), (3) Low data cost (perfect for small, periodic transmissions), (4) Nationwide coverage through cellular networks, (5) Secure, licensed spectrum communication.

NB-IoT connectivity Decade-long battery life Bank-grade security Utility-scale reliability

Elephant Intrusion Alert System

Life-Critical Public Safety (Sagaj / Prixmation)

A government-recognized wildlife tracking and early warning system protecting over 200 villages in elephant corridors. Human-elephant conflict kills hundreds annually—this system provides advance warning when wild elephants approach settlements, giving communities time to evacuate safely and protect their crops and homes.

GPS Tracking Collars Machine Learning (TensorFlow) GSM/4G Edge Siren Units Mobile Alert System Solar Powered
Wildlife Tracking & Alert APIs
  • Real-time GPS tracking API with 2-minute update intervals
  • Geofencing API with dynamic village boundary management
  • ML prediction API estimating elephant trajectory and arrival time
  • Multi-channel alert API (SMS, mobile push, siren activation)
  • Historical movement API for corridor mapping and pattern analysis
  • Government dashboard API for forest department monitoring

System Components

  • GPS Tracking Collars: Ruggedized wildlife collars on key elephants (herd leaders, lone males). Solar-charged, waterproof, reporting location every 2 minutes during movement, every 30 minutes when stationary.
  • ML Movement Analysis: TensorFlow model trained on 3 years of elephant movement data predicts trajectory, speed, and likely destinations with 95% accuracy
  • Edge Siren Units: Solar-powered sirens installed at village peripheries. Activate automatically when elephants detected within 2km radius, providing 15-30 minute advance warning
  • Mobile Alert System: SMS and app-based alerts sent to registered villagers, forest guards, and village leaders with elephant location, distance, and estimated arrival time
  • Government Integration: Real-time dashboard for forest department enables rapid response team deployment and coordination

Intelligence Layer

Direction Prediction: System analyzes last 10 GPS points to determine heading and speed. Elephants moving at 3 km/h toward village trigger alerts; elephants at same distance but moving away do not.

Speed Analysis: Slow movement (0.5-1 km/h) = foraging behavior, low threat. Fast movement (4-6 km/h) = traveling, high threat if heading toward settlement. System adjusts alert urgency accordingly.

False Positive Suppression: Elephants often wander near villages without entering. System requires 3 consecutive position updates showing approach trajectory before activating sirens. This prevents alert fatigue while maintaining safety margins.

Saves human lives Protects wildlife Government recognized Media coverage

Smart Remote Healthcare System

Preventive, Continuous Care for Underserved Areas

A telemedicine platform bringing hospital-grade health monitoring to rural and remote areas lacking medical infrastructure. By continuously monitoring vital signs and detecting early warning signs of health deterioration, this system enables preventive intervention before conditions become critical, reducing emergency hospitalizations and improving health outcomes.

ESP32 AD8232 ECG MAX30102 (SpO₂/HR) Blood Pressure Module Glucose Sensor Cloud Analytics
Healthcare Data APIs
  • Vital signs streaming API with real-time telemetry
  • ECG analysis API with arrhythmia detection algorithms
  • Trend analysis API identifying gradual health deterioration
  • Doctor dashboard API with patient history and alerts
  • Emergency alert API with severity-based routing to local clinics
  • Prescription reminder API integrated with mobile app

Parameters Monitored

  • ECG (Electrocardiogram): Continuous heart rhythm monitoring detecting arrhythmias, irregular beats, and cardiac events in real-time
  • Heart Rate: Resting HR, exercise HR, HR variability (HRV) as indicator of autonomic nervous system health
  • SpO₂ (Blood Oxygen): Critical for respiratory conditions, COPD, pneumonia detection. Levels below 90% trigger immediate alerts
  • Blood Pressure: Automated cuff measurements 3x daily, tracking hypertension control and medication effectiveness
  • Blood Glucose: Fingerstick readings for diabetic patients, trend analysis for insulin adjustment recommendations
  • Respiratory Rate: Breaths per minute calculated from ECG or SpO₂ waveform, early indicator of respiratory distress
  • Body Temperature: Fever detection and infection monitoring

Clinical Intelligence & Early Warning

The system doesn't just record data—it interprets it. Machine learning models trained on clinical datasets identify patterns indicating health deterioration hours to days before symptoms become severe.

Trend Analysis Examples: Gradual SpO₂ decline over 3 days (96% → 93% → 90%) suggests worsening respiratory condition—alert sent before acute crisis. Increasing resting heart rate + decreasing HRV + rising BP = early signs of cardiovascular stress or infection.

Brings healthcare to rural areas Early diagnosis & intervention Reduces hospitalizations Secure medical-grade system

Engineering Principles

Edge-First Logic

Intelligence at the edge, not dependent on cloud connectivity. Systems make decisions locally with real-time sensor data.

Offline Safety

Systems continue operating safely even when internet fails. Local decision engines ensure continuity of operations.

Sensor Fault Detection

Continuous validation and sanity checking of sensor data. Multi-sensor fusion prevents single points of failure.

Modular Scalability

Systems designed to scale from single units to large deployments without architectural changes.

Get In Touch

Let's build reliable, intelligent systems together.