AI / ML Engineer
Core developer of the no-code agent platform MIA
- Next-gen AI agents for MIA.
- RAG-2.0 pipeline → smarter chunking, hybrid BM25 + vector rerankers, latency < 1000 ms for voice/chat.
- Secure API gateway with key-scoped roles, rate-limits, and usage analytics (FastAPI + PostgreSQL).
- Managed metadata DB for agent configs, session traces, and audit logs.
- Outbound call scheduling + Twilio Integration: Working on a productioon grade soltuion for initiaing outbound call according to schedule with logging and tracing
- Bot orchestration @ MIA – FastAPI + async-Python layer sustaining 100 + concurrent sessions with 99.9 % uptime.
- Multi-LLM router – dynamic prompt & function-call pipelines across OpenAI, Claude, Gemini, LLaMA.
- RAG ingestion service – vectorizes PDFs/DOCX/CSV/XLSX, stores in FAISS, retrieves context in < 3000 ms.
- Twilio SIP ↔ Zoom bridges – real-time voice calls, live ASR/TTS, S3 call recordings, auto-transcripts.
- QuantumDataLytica “machines” – pluggable components for workflow-automation marketplace.
- CV + OCR journey-mapping – solo project converting shopping screen-recordings into structured datasets.
Python |
OpenCV |
Scikit-Learn |
TensorFlow |
PyTorch |
Git |
OpenAI |
LangChain |
LangGraph |
Livekit |
PostgreSQL |
| Year | Milestone |
|---|---|
| 2025 | Scaled MIA orchestration layer to 100 + live sessions with < 1 % error rate |
| 2025 | Integrated Twilio SIP & Zoom for seamless telephony/meeting joins |
| 2025 | Co-designed and implemented a dynamic voicebot configuration system, enabling flexible deployment and real-time usage management. |
| 2025 | Implemented a dynamic MIA chatbot WebSocket framework, enabling real-time configuration and management of chatbot sessions. |
| 2024 | Shipped CV + OCR pipeline mapping consumer shopping journeys |
