I am a B.Tech Computer Science & Engineering student at MIT ADT University, Pune, focused on building AI/ML applications, backend systems, developer tools, data-intensive platforms, and production software.
My work sits at the intersection of:
Artificial Intelligence × Backend Engineering × Data Systems × Software Reliability × FinTech
I am especially interested in systems where the difficult problems are not the happy path, but:
uncertainty · bad data · failure · inconsistent state · unreliable networks · deployment risk
|
|
|
|
|
Evidence & Trust Control Plane
TrustMesh is a local-first, pre-transaction evidence and trust system designed to help evaluate transactions, payment destinations, identities, and claims before action. Instead of reducing trust to a single binary prediction, TrustMesh organizes evidence, preserves provenance, evaluates consistency, represents uncertainty, and identifies what evidence should be checked next.
|
Pre-Deployment Software Risk Analysis
PreFlight is a developer-focused system designed to analyze the risk introduced by software changes before production deployment. Instead of: PreFlight thinks in terms of:
|
| SYSTEM | PROBLEM | CORE IDEA | DECISION |
|---|---|---|---|
| 🕸️ TrustMesh | Trust under uncertainty | Evidence + reasoning | PASS / VERIFY / STOP |
| 🛫 PreFlight | Deployment under uncertainty | Impact + failure analysis | SAFE / CAUTION / DO NOT DEPLOY |
Identify uncertainty → gather evidence → reason about risk → make the decision explainable.
A full-stack logistics platform built for an actual transport business.
The project moved my engineering thinking from:
"Can I build it?"
to:
"Can someone depend on it?"
TRIPS
├── Revenue
├── Collections
├── Realtime Updates
├── Offline Submission
├── Field Workflows
└── Mobile Deployment
React 19 TypeScript Node.js MUI Capacitor Server-Sent Events
Unreliable Connectivity · Realtime State · Offline Queueing · Field Usage · Operational Data Integrity
Status: 🟢 Production
A systems-design project focused on financial data integrity, auditability, statutory compliance, and explainable payroll processing.
The architecture treats payroll as a controlled financial process rather than a simple salary calculator.
PAYROLL INPUT
↓
RULE VERSIONING
↓
STATUTORY CALCULATION
↓
IMMUTABLE SNAPSHOT
↓
FINANCIAL LEDGER
↓
AUDIT TRAIL
- Immutable payroll snapshots
- Versioned PF / ESI / PT rules
- Explainable deductions
- Append-only financial ledger
- Idempotent payroll execution
- Audit-oriented data modeling
FastAPI PostgreSQL React TypeScript Docker
Status: 🟡 Architecture + Development in progress
|
YOLOv8 OpenCV MediaPipe |
scikit-learn TensorFlow |
Hugging Face LLM APIs |
Python FastAPI |
🧩 SchemaSense AI — Semantic JSON Inference Platform
Hybrid JSON intelligence platform combining deterministic validation with ML-assisted semantic type inference.
Capabilities
- Malformed JSON detection and repair
- Semantic type inference
- Context-aware TypeScript generation
- ML-assisted prediction
- Backend API integration
Stack: React TypeScript Node.js Express FastAPI scikit-learn
Repository: https://github.com/srthck/schemasense-ai
🚑 AmbuFlow — Emergency Ambulance Coordination
Emergency coordination platform with authentication, role-based workflows, dashboards and live tracking.
Stack: Node.js Express MongoDB Firebase Google Maps
Repository: https://github.com/srthck/AmbuFlow
👁️ YOLOv8 Acne Analysis — Computer Vision
Computer-vision pipeline for acne detection and classification using YOLOv8 and image-processing techniques.
Stack: Python YOLOv8 OpenCV
Repository: https://github.com/srthck/yolov8-acne-analysis
🖼️ VisionCaption AI — Deep Learning
Computer-vision and deep-learning application for generating natural-language descriptions from images.
Stack: Python TensorFlow OpenCV
| Languages | Python · Java · C++ · TypeScript · JavaScript · SQL · Kotlin |
| Backend | FastAPI · Node.js · Express · REST APIs |
| Frontend | React · Next.js · TypeScript · Tailwind CSS |
| Mobile | Android · Kotlin · Jetpack Compose · React Native · Capacitor |
| Databases | PostgreSQL · MongoDB · MySQL · SQLite · Firebase |
| AI / ML | scikit-learn · TensorFlow · YOLOv8 · OpenCV · MediaPipe · Hugging Face |
| Infrastructure | Docker · Git · GitHub Actions · Linux |
| Engineering | System Design · API Design · Data Modeling · Authentication · Authorization · Idempotency · Audit Trails · Offline-First Architecture · Realtime Systems · Risk Analysis · Failure Handling · Observability |
Worked across multiple full-stack applications involving:
- JWT authentication and authorization
- Employee management systems
- React / Node.js / MongoDB applications
- Realtime communication
- Financial-management workflows
Recognition: Letter of Recommendation
React Node.js JavaScript MongoDB
Completed a one-month full-stack development internship delivering application assignments across modern web technologies.
Recognition: Certificate of Completion + Letter of Recommendation
React Node.js JavaScript APIs
| DATA Model state before UI |
FAILURE Design beyond the happy path |
TRUST Preserve evidence + provenance |
RISK Make uncertainty explicit |
| INTEGRITY Protect system state |
AUDIT Make decisions reconstructable |
IDEMPOTENCY Make repetition safe |
DEPLOY Production is part of engineering |
01 Understand the real problem
02 Model the system and data
03 Define boundaries and invariants
04 Design APIs and core architecture
05 Build the backend
06 Integrate AI where it adds real value
07 Test failure cases
08 Deploy
09 Observe
10 Iterate
→ What if the input is malformed?
→ What if the network disappears?
→ What if a dependency fails?
→ What if the same request runs twice?
→ What if state becomes inconsistent?
→ What if the model is uncertain?
→ What if production behaves differently?
→ Can the system explain what happened?
|
|
|
|
| Status | Project | Engineering Signal |
|---|---|---|
| 🟢 | Shivam Transport | Production / Real-world Operations |
| 🟢 | SchemaSense AI | AI + Backend + Semantic Inference |
| 🟢 | AmbuFlow | Realtime / Full Stack |
| 🟢 | YOLOv8 Acne Analysis | Computer Vision |
| 🟢 | VisionCaption AI | Deep Learning |
| 🔵 | TrustMesh | Trust + AI + Evidence Systems |
| 🔵 | PreFlight | Developer Infrastructure + Risk |
| 🟡 | MSME Payroll | FinTech + Auditability |
AI Engineer · ML Engineer · Backend Engineer · Software Engineer · Python Developer · FastAPI Developer · Full Stack Developer · Computer Vision Engineer · Data Engineer · FinTech Engineer · Systems Engineer · Developer Tools · AI Infrastructure · Software Reliability · System Design · API Development · PostgreSQL · Docker
```
Also, I removed the old trophy section because it was visual noise, while retaining the useful animated elements from your original profile: header, typing effect, stats, streak, contribution graph, and project presentation. Your original README already used those visual mechanisms.
After pasting: click Preview. The <table>, <details>, headings, badges, and diagrams should now render as UI instead of showing the source code.
And yes: Payroll is explicitly IN DEVELOPMENT throughout this version.