Reinforcement-learning service for self-optimizing 5G/6G networks. I built the MRO module end-to-end and re-architected the training pipeline from linear to multi-threaded, cutting runtime by 45–55%. Unified RL harness across MRO, CCO, Energy Saving, and Load Balancing rApps.
Menon Pranto
AI/ML Engineer · O-RAN & 5G Networks
"""I build self-optimizing 5G networks with reinforcement learning."""
role = "AI/ML Engineer @ Cloudly IO"
work = "core contributor, Maveric (Linux Foundation)"
focus = ["reinforcement learning", "Bayesian digital twins", "O-RAN"]
open_to = "AI/ML engineer roles — research track"
location = "Dhaka, Bangladesh (UTC+6)"AI/ML Engineer at Cloudly IO and core contributor to Maveric (Linux Foundation Connectivity). I build production reinforcement-learning agents and Bayesian digital twins for AI-native 5G/6G networks — including the end-to-end PPO-based MRO pipeline, which I re-architected from linear to multi-threaded for a 45–55% runtime reduction, and a vectorized handover protocol that runs 5.5× faster on ARM and 18× on Intel. Started in data annotation and QA, learned what good training data looks like from the ground up, and now ship end-to-end ML pipelines from data sim through xApp/rApp deployment.
projects
Golden topology generator powering Maveric's rApps. I wrote the 3GPP-compliant golden generators — Hata/UMa/UMi path loss, Gauss-Markov mobility across four UE velocity classes, and ECI computation.
GPyTorch Gaussian-process engine for RF propagation prediction. I hardened the serving layer: 3-tier cache (Redis → MongoDB → Postgres), partition-aware exact-offset Kafka commits, and Prometheus-instrumented training histograms.
RL-based adaptive firewall with real-time malware detection. Learns from network activity and shares collaborative threat intelligence across devices — ML models and cloud integration designed for fast, device-specific protection.
Seven-class facial-emotion classifier fine-tuned on FER2013 — ResNet backbone with class-weighted loss for imbalance, test-time augmentation, and multi-GPU training over NCCL.
Zero-dependency PWA for volleyball referees and scorekeepers. Tracks scores, rotations, substitutions, timeouts, and libero usage in real time with automatic rule enforcement. Vanilla JavaScript.
experience
skills
# languages
- Python
- C/C++
- Java
- JavaScript
- SQL
# ML & RL
- PyTorch
- TensorFlow
- scikit-learn
- Stable-Baselines3
- GPyTorch
- NumPy
- Pandas
# LLM & agents
- LangGraph
- LangChain
- MCP / fastmcp
- pgvector
- RAG
- Claude API
# infra & data
- FastAPI
- Kafka
- Docker
- Kubernetes
- AWS (EKS · S3 · Cognito)
- PostgreSQL
- Redis
- MongoDB
- Prometheus
- OpenTelemetry
# domain
- O-RAN
- 3GPP
- Bayesian digital twins
- RL / PPO
- xApp / rApp