AI & Telecom Platform Engineer · Nybsys

I build the software that manages mobile networks, the AI that helps operate them, and the platforms they run on.

Network management systems for 4G LTE small-cell fleets, LLM agents that reason over live RAN KPIs and alarms, and the Kubernetes platforms that carry both into operator networks.

See my work Get in touch

years shipping telecom & AI products: 4+
products shipped, from PTT to RAN AI: 5
operator & partner deployments / demos: 3

ran-ops-agent

MCP

Why did throughput drop at this site overnight?
tool performance · last 24 h time-series
tool alarms · same site · same window analytics
tool knowledge · related guidance RAG
verified against 3 sources
Likely root cause: a radio alarm on one cell after midnight, matching the throughput dip. Both cleared after an early-morning reset.

Illustrative session · the kind of agent I build

Deployed & demoed with

Focus areas

Three areas, one stack

Most of my projects touch all three. Pick one to filter the case studies.

Applied AI & LLM Agents

RAN-ops agents with root-cause analysis and RAG knowledge bases, MCP tool servers, and fully offline voice AI on edge GPUs.

2 projects

Telecom & Network Management

TR-069 ACS, SNMPv3, fault/performance/config management and zero-touch provisioning for small-cell RAN fleets.

4 projects

Kubernetes & Platform

Production Kubernetes for operator deployments: networking, TLS ingress, monitoring and observability.

5 projects

Selected work

Case studies

Products I have built and run at Nybsys, from 2022 to today.

AI Telecom Platform

Featured

Gen-AI RAN Operations Agent

Apr 2026 – Jul 2026 · Root-cause analysis · RAG · MCP tools · UI

An LLM agent that answers engineers’ questions about live radio-network performance and alarms, finds root causes, and runs scheduled work.

Root-cause analysis: investigates network problems across performance and alarm data, and explains the likely cause, not just the symptom.
Built the MCP tool servers that give the agent access to RAN performance and alarm data.
Moved performance data to a time-series database and alarms to a column store, both suited to fast analytical queries.

Python
MCP
LLM agents
RAG
InfluxDB
ClickHouse
PostgreSQL
React
+1

Telecom Platform

Featured

Hermes NMS

Aug 2026 – present · NMS · TR-069 · ZTP · current work

A network management system for small-cell fleets, built on my own TR-069 implementation.

Wrote the TR-069 (CWMP) protocol implementation that devices connect to. Provisioning and configuration management are built on it.
Brought zero-touch provisioning to a working end-to-end state, so new devices configure themselves when they come online.
Built configuration pushes to many devices in parallel, with bulk rollback and progress tracking that stays visible in the UI.

Python
TR-069 / CWMP
PostgreSQL
ZTP

Telecom Platform

Featured

Nexus Insight NMS

Oct 2024 – Mar 2026 · Flagship NMS · operator deployments

A cloud-native network management system for small-cell fleets: device, fault, performance, and configuration management with northbound integration.

Built a TR-069 auto-configuration server in Python, with device connectivity tooling and firmware upgrades.
Wrote the Kubernetes deployments, staged and tested them, and documented how to build the cluster.
Scaled file transfer and device data storage, and started load testing the auto-configuration server.

Python (asyncio)
TR-069 / CWMP
SNMPv3
MongoDB
InfluxDB
RabbitMQ
Redis
Kubernetes
+4

AI Platform Telecom

Sentra V2: AI-Enabled Push-to-Talk

Jun 2023 – Mar 2025 · MWC 2025 demos with Intel & Lanner

A PTT platform whose channels include an on-premise AI agent that hears, sees, and answers, with no internet connection.

Python
Open-weight LLMs
RAG
Multimodal
Speech-to-text
+5

Platform

Sentra PTT: Platform Foundations

Aug 2022 – Jun 2023 · Docker → Kubernetes

Containerizing the original Sentra server and putting it on Kubernetes.

Docker
Kubernetes
Nginx
Prometheus

Working on AI for network operations, telecom OSS or Kubernetes platforms?

Let’s talk

Skills

Skills, with the receipts

Every skill below is tied to work I shipped. Open one to see where it came from.

AI & LLM Systems 8

MCP tool servers 1
- Tool servers giving an agent access to RAN performance and alarm data Gen-AI RAN Operations Agent
LLM agents 2
- RAN operations agent with root-cause analysis, scheduled work and extensible skills Gen-AI RAN Operations Agent
- Voice AI agent inside PTT channels, with conversational memory Sentra V2: AI-Enabled Push-to-Talk
Root-cause analysis 1
- Agent that explains the likely cause of network problems from performance and alarm data Gen-AI RAN Operations Agent
Answer verification 1
- A second model that checks agent answers before users see them Gen-AI RAN Operations Agent
RAG 2
- Knowledge base that grounds the RAN agent’s answers Gen-AI RAN Operations Agent
- Retrieval over user-uploaded documents, running on a local GPU Sentra V2: AI-Enabled Push-to-Talk
Multimodal models 1
- Questions that combine a photo and text Sentra V2: AI-Enabled Push-to-Talk
On-prem GPU inference 2
- Full voice + LLM pipeline on a local GPU with no internet dependency Sentra V2: AI-Enabled Push-to-Talk
- MWC 2025 demos on edge GPU hardware Sentra V2: AI-Enabled Push-to-Talk
Speech AI (STT / TTS) 1
- Offline speech pipeline with improved voice quality Sentra V2: AI-Enabled Push-to-Talk

Telecom & OSS 7

TR-069 / CWMP 2
- Wrote the TR-069 protocol layer for Hermes NMS Hermes NMS
- TR-069 auto-configuration server and firmware upgrades Nexus Insight NMS
SNMPv3 1
- SNMPv3 fault traps for northbound systems, validated on real radios Nexus Insight NMS
FM / PM / CM 1
- Fault, performance and configuration management with scheduled exports and reports Nexus Insight NMS
Zero-touch provisioning 1
- End-to-end ZTP with model-safe configuration Hermes NMS
Fleet configuration 2
- Bulk configuration for groups of sites Nexus Insight NMS
- Parallel pushes, bulk rollback and task reports Hermes NMS
RAN KPIs & alarms 2
- Performance and alarm data made queryable by an LLM agent Gen-AI RAN Operations Agent
- Alarm synchronization and device health alarms Nexus Insight NMS
Push-to-talk 1
- User, group, channel and access management for a PTT server Sentra V2: AI-Enabled Push-to-Talk

Kubernetes & Platform 7

Kubernetes 3
- Production deployments and cluster-build documentation for an NMS Nexus Insight NMS
- Kubernetes setup for Sentra V2 Sentra V2: AI-Enabled Push-to-Talk
- First containerization and Kubernetes move for Sentra Sentra PTT: Platform Foundations
Kubernetes networking 1
- Multi-network pods and cluster DNS for operator environments Nexus Insight NMS
Ingress & TLS 2
- TLS-terminated web portal behind a reverse proxy Nexus Insight NMS
- Reliable delivery of long-running agent tasks Gen-AI RAN Operations Agent
Observability 1
- Device monitoring with Prometheus; Grafana/Loki log dashboards (in progress) Nexus Insight NMS
Docker 2
- Containerized services and tooling Nexus Insight NMS
- Containerized PTT server with multiple instances Sentra PTT: Platform Foundations
Operator networks 1
- Deployments and troubleshooting inside operator networks Nexus Insight NMS
Load testing & simulation 1
- Base-station traffic simulator; server load testing (in progress) Nexus Insight NMS

Backend & Data 7

Python (asyncio) 2
- Async network-management services Nexus Insight NMS
- TR-069 protocol implementation Hermes NMS
InfluxDB 2
- Time-series store for RAN performance data Gen-AI RAN Operations Agent
- Performance data storage Nexus Insight NMS
ClickHouse 1
- Analytical store for alarm data Gen-AI RAN Operations Agent
PostgreSQL 2
- Configuration and task data Hermes NMS
- Migration to purpose-built stores Gen-AI RAN Operations Agent
MongoDB 1
- Scaled device and fault data storage Nexus Insight NMS
RabbitMQ & Redis 1
- Resilient messaging and task handling Nexus Insight NMS
File pipelines 1
- Scaled file transfer and northbound data delivery Nexus Insight NMS

Frontend 2

React 1
- Production agent UI with sign-in and transparent answers Gen-AI RAN Operations Agent
Operations UIs 2
- Task status, KPI dashboard and device views Hermes NMS
- Network map and management portal Nexus Insight NMS

About

Hi, I’m Mahmud.

Since 2022 I have worked on Nybsys products end to end, from push-to-talk servers to network management systems. More recently I have been building generative-AI agents for network operations.

My work sits where three areas meet. One is telecom protocols and OSS (TR-069, SNMP, FM/PM/CM). Another is applied AI (RAG, multimodal models, MCP tool servers, answer verification). The third is the Kubernetes platform work that ships both to operators such as GP.

Timeline

2026
Hermes NMS
Aug 2026 – present
2026
Gen-AI RAN Operations Agent
Apr 2026 – Jul 2026
2024
Nexus Insight NMS
Oct 2024 – Mar 2026
2023
Sentra V2: AI-Enabled Push-to-Talk
Jun 2023 – Mar 2025
2022
Sentra PTT: Platform Foundations
Aug 2022 – Jun 2023

Contact

Let’s build software that keeps networks running.

I’m interested in AI for network operations, telecom OSS and Kubernetes platform work.