I'm a backend engineer based in New Delhi, building production systems at the intersection of Python APIs, cloud infrastructure, and ML-integrated products. Currently shipping identity & fintech platforms at U2opia — and always open to remote work and freelance projects on the side.
I care about systems that actually work in production: fast, compliant, and maintainable. Less buzzword soup, more shipped code.
FastAPI / Spring Boot backends → REST APIs, auth flows, data pipelines
Identity & KYC platforms → Face liveness, document verification, multi-region compliance
ML-integrated products → Anomaly detection, computer vision, model deployment
IoT + Edge systems → Sensor pipelines, embedded devices, cloud connectivity
AWS infrastructure → Rekognition, S3, EC2, Lambda
FastAPI · AWS Rekognition · React · Multi-region compliance
End-to-end KYC platform I led from zero to production, targeting India and UAE. Real-time face liveness detection, identity document verification, and onboarding workflows — all under 300ms p95 latency. Reduced manual onboarding effort by ~60% for client ops teams. Shipped from MVP to go-live in 3 months.
Java · Spring Boot · MySQL
Designed and shipped the full charging, subscription, and unsubscription flow for a telecom value-added services platform. Zero-defect launch in production.
Python · CI/CD · 13+ repositories
Owned the full migration of a live gaming platform across 13 repos — restored CI reliability and unblocked deployments within a 6-week window.
| Project | What it does | Tech |
|---|---|---|
| OncoVision | Breast cancer classification — 96% accuracy with confidence scoring | Scikit-Learn, XGBoost, SVM |
| Home Gesturize | Real-time gesture-controlled home automation, <100ms latency | ESP32, MediaPipe, KNN, Python |
| Coffee Roasting Quality Prediction | Neural net trained on chemical composition data to grade roast quality | TensorFlow, Python |
An end-to-end IoT pipeline for predictive maintenance — the kind of thing that would actually run in a factory or remote site.
Setup:
- Edge device: Raspberry Pi 4
- Sensor: DHT22 temperature & humidity sensor (2-wire: data pin → GPIO4, VCC → 3.3V, GND → GND — wiring is dead simple, no extra components needed)
- Cellular connectivity: Quectel EVB (EC25) for sending data over 4G when Wi-Fi isn't available
- Language: Python throughout
What it does (in progress):
- Read live temperature/humidity data on the Pi
- Basic threshold-based anomaly detection (flag readings outside normal range)
- Send alerts + data to a cloud API via the Quectel cellular module (AT commands over serial)
- Store time-series data and visualize trends on a dashboard
- Run a lightweight ML model (Isolation Forest or similar) on the edge for anomaly scoring
The goal is a minimal but realistic predictive maintenance loop — sense → process → transmit → visualize — built with off-the-shelf hardware and clean Python code. Will be open-sourced with full wiring diagrams and setup docs when stable.
Backend: Python (FastAPI, Django DRF) · Java (Spring Boot) · REST APIs · MQTT
ML / AI: TensorFlow · PyTorch · Scikit-Learn · OpenCV · NLP · Computer Vision
Cloud & Infra: AWS (Rekognition, S3, EC2, Lambda) · Docker · Linux
Databases: MySQL · Redis · SQLite3 · MariaDB
Embedded & IoT: Raspberry Pi · Arduino · ESP32 · NodeMCU · Renesas · MQTT
Frontend: React.js · TypeScript · HTML/CSS
Tools: Postman · Git · Streamlit · Selenium
- 📄 Published at ICCCESB 2025 — Cognitive Intelligence and Big Data: A Symbiotic Approach to Predictive Analytics in Healthcare
- 🎓 B.Tech CSE (Machine Learning), Gautam Buddha University — CGPA 9.43
- 🌍 Built systems running live across India, UAE, Poland, and Africa
I'm open to remote roles and freelance projects — especially anything involving backend APIs, ML integration, cloud infrastructure, or IoT pipelines.
📧 singh.parnika07@gmail.com 💼 linkedin.com/in/parnika-singh07 🧩 leetcode.com/u/parnika_singh07