"I don't just build demos; I architect production systems that handle scale, concurrency, and hallucinations."
- 🎤 Real-Time Voice AI: Built full-duplex agents handling 1,000+ concurrent connections with <150ms latency.
- 🧠 RAG at Scale: Engineered retrieval for 50,000+ documents achieving 92% accuracy via semantic chunking.
- ⚙️ Backend Reliability: Production uptime of 99.8% using Docker, CI/CD, and robust error handling.
| Category | Technologies |
|---|---|
| Generative AI |     |
| Backend & API |     |
| DevOps & Cloud |     |
| Vector DBs |    |
Full-duplex voice AI with sub-second latency.
- Tech: FastAPI, Deepgram, WebSockets, Ragas.
- Impact: Handles 1,000+ concurrents, <150ms latency, custom hallucination filter.
- Signal: 🟢 Production Ready
Enterprise-grade retrieval system for massive datasets.
- Tech: LangChain, Gemini, Semantic Chunking, ChromaDB.
- Impact: 92% retrieval accuracy, processed 50k+ docs, reduced query time by 90%.
- Signal: 🟢 High Scalability
Complex PDF generation pipeline using LLMs.
- Tech: Python, Prompt Chaining, Jinja2.
- Impact: cut manual work by 95% (45 min → 90 sec), 100% data accuracy.
- Signal: 🟢 Business Automation
No-code automation infrastructure.
- Tech: n8n, Webhooks, API Integrations.
- Impact: Automated 87% of manual goals, seamless data sync.
- Signal: 🟢 DevOps
"Talk is cheap. Show me the code."