Software engineer working across TypeScript, Python, and applied machine learning. I care about small, well-tested pieces of software: libraries with clear boundaries, agents that do one job reliably, and models that earn their complexity.
Currently building and publishing reusable TypeScript packages, and exploring agentic AI patterns and LLM orchestration.
💼 5 Public Projects | 📦 Published on npm | ⭐ Open to Collaborate
Calendar dates without a timezone — published on npm
- A year, a month, a day, and the arithmetic that goes with them
- Zero-dependency, fully typed, 33 unit tests green in CI
npm install @dyadav0607/calendardate- Tech: TypeScript, Vitest, npm
Production-grade autonomous agent using ReAct pattern for content generation
- Multi-LLM orchestration (OpenAI GPT-4o, Claude Sonnet 4)
- Autonomous research → write → review → publish workflow
- Self-refinement loop with 8.0/10 quality threshold
- GitHub Actions automation (generates 30 articles/month)
- Tech: Python, ReAct Pattern, LLM APIs, CI/CD
- calendardate - Timezone-free date arithmetic, published on npm
- MediumAgent - Agentic AI with ReAct pattern & multi-LLM orchestration
- Baby Safety Detection - Deep learning for safety monitoring
- Anomaly Detection in Time Series - Isolation forest + DeepAnT; HackIllinois 2020 winner
- POS Tagging with HMM - Statistical NLP
- 📦 calendardate - Publishing a zero-dependency date library to npm
- 🤖 MediumAgent - Autonomous agentic AI publishing articles via ReAct pattern
- 🧠 Exploring agentic AI patterns and LLM orchestration
- 📚 Extracting reusable libraries out of production TypeScript work
💡 Always learning, always building