Backend-focused Full-Stack Developer | AI/ML Engineer
Building production-grade ML systems, LLM applications, and intelligent automation from India. I believe in creating systems that are simple to operate and easy to delete.
- LLMs & AI Systems: RAG pipelines, memory-augmented agentic systems, multi-agent orchestration, context-aware chatbots
- ML Engineering: Model optimization, inference latency reduction, monitoring/evaluation pipelines
- Computer Vision: OCR, object detection, image preprocessing, segmentation, post-processing pipelines
- Data Generation: Synthetic data generation with SDV and CTGAN, tabular modeling
- Security & Fraud: Fraud detection, biometric verification, deepfake/catfish detection
- Healthcare AI: OCR/PDF parsing, structured extraction, disease prediction workflows
- MLOps: End-to-end deployment with Docker, RunPod, API integrations, cloud tooling
- Backend services with Node.js, Express, and MongoDB
- System design — focusing on scalability and data modeling
- Active open-source contributions in AI and developer tooling
| Languages |
|
| Frontend |
|
| Backend |
|
| Databases |
|
| DevOps & Cloud |
|
| AI/ML & Data Science |
|
| Message Queues |
|
| Monitoring |
|
| Tools & Others |
|
InsForge — Backend-as-a-Service Platform
- Implemented multi-admin support for self-hosted dashboard
- Replaced single env-var admin with database-backed admin model
- Added bcrypt password hashing, soft-delete, timing-attack–resistant credential checks
- Implemented root-only management endpoints
- Merged into
feat/multi-adminintegration branch
Novu — Open-source notification infrastructure
- Active contributor to the notification platform
- Working on notification delivery optimization and workflow improvements
- Contributing to the notification feed management system
RoboCurve — Open-source robotics evaluation platform
- Contributing to simulation and control systems
- Working on integration with reinforcement learning frameworks
- Enhancing simulation accuracy and performance
Tracer — Open-source AI inference research system
- Contributing to adaptive inference and coordinated intelligence systems
- Exploring model coordination, inference efficiency, and evaluation
- Working with research-oriented AI infrastructure and open-weight models
Superlog — Open-source AI observability and automated debugging
- Contributing to AI-powered observability and automated bug-fixing workflows
- Working with logs, traces, metrics, OpenTelemetry, and incident analysis
- Exploring automated PR generation, debugging agents, and developer tooling
JSON Schema — JSON Schema specification and tooling
- Contributing to JSON Schema validation and tooling
- Working on schema validation improvements and documentation
- Community engagement and specification discussions
| Project | Description | Tech Stack | Links |
|---|---|---|---|
| Graphsight Pro | Local-first RAG observability engine tracking "Retrieved vs. Used" context gaps using zero-cost lexical trace algorithms to eliminate token bloat | Python, LangGraph, Pydantic, FastAPI, React | GitHub · Live Demo · PyPI |
| CineSpot | 3D interactive cinema ticket booking platform with realistic auditorium sightline simulation and zero-double-booking guarantees using PostgreSQL pessimistic row locking | React, TypeScript, Three.js, Node.js, PostgreSQL | GitHub · Live App |
| PingGod | Distributed website uptime monitoring system with Redis Stream worker queues, incident state machines, and Next.js status pages | TypeScript, Next.js 15, Express, Redis Streams, PostgreSQL | GitHub · Live API |
| LetAgents | Cloud platform providing autonomous coding agents with infrastructure provisioning, deployment, and operational observability | Next.js, TypeScript, Drizzle ORM, PostgreSQL, Tailwind | GitHub · Live Site |
| Canteeno | High-concurrency backend ordering system with JWT authentication, WebSocket real-time updates, and optimized MongoDB queries | Node.js, Express, MongoDB, WebSocket, Redis | GitHub · Live App |
- Portfolio: amreshdev.me
- LinkedIn: in/amresh-chaurasiya
- X (Twitter): @Amresh__01
- Email: amresh.codes@gmail.com
- GitHub: github.com/Amresh-01
"Most of what's interesting is in the commits, not the README."