#blogs
Caching TTS Audio with Content Hashing
TTS
Voice Agents
Caching
Python
Synthesizing the same sentence twice is wasted latency and money. A practical guide to deterministic cache keys, storage layouts, and the pitfalls that silently break TTS reuse.
Anatomy of a Kamailio Config
Kamailio
SIP
VoIP
Telephony
A walk through a production-style kamailio.cfg — max-forward checks, sanity validation, in-dialog routing, CANCEL handling, and ACLs, explained line by line.
#projects
GoldenGraphRAG
Python
PostgreSQL
Apache AGE
openCypher
Docker
Pipeline that turns text datasets into queryable knowledge graphs — documents are chunked, entities and relationships extracted under a YAML-defined ontology, deduplicated, and stored as a property graph in PostgreSQL via Apache AGE, queried with openCypher. Works with any OpenAI-compatible LLM provider.
Content Style Transfer using RAG
Python
Next.js
Gemini API
RAG
Docker
LLM-as-a-Judge
Retrieval-augmented pipeline that rewrites email content in a different style while preserving meaning. Ingests .eml files in bulk or via filesystem watching, retrieves similar emails for context, generates with Gemini or any Ollama/OpenAI-compatible endpoint, and scores results with an LLM-as-a-judge evaluation loop.
Language-Learner
Python
SQLite
Docker
German vocabulary learning app built on a database of the 10,000 most common words, with a Python backend, web frontend, and Docker packaging.
#contacts
I'm interested in working on ambitious projects with positive people and goal-oriented companies.
However, if you have other request or question, don’t hesitate to contact me.
Reach out