Deep & Proven Technology knowledge across entire architecture of the solutions proposed in AI and Automation in different IT/Enterprise/Telco use cases with values brought to organization.

Use Cases lead by me:

• Architected Root Cause Analysis (RCA) AI Agent, which showcased autonomous reasoning to identify the 5G network problem it received from the alerting system and providing a recommendation to fix it. Showed the Orchestrating Agent managing the entire workflow including invoking other agents and MCP servers.
• Architected retrieval-augmented generation (RAG) pipelines using LangChain, OpenAI embedding, and vector databases (FAISS, Pinecone) to enable intelligent RCA summaries from BSS logs and 5G CNF alerts.
• Developed predictive AI and anomaly detection frameworks using time-series models (LSTM), Isolation Forests, and Databricks MosaicAI to monitor KPIs across FCAPS domains in OSS/BSS. Integrated MLOps practices for training, retraining, and model lifecycle management.
• Built LLM-enabled retrieval and recommendation pipelines using LangChain, FAISS, Pinecone, and Weaviate to augment post-anomaly workflows by retrieving relevant KB articles, configuration history, and RCA documentation to assist NOC and DevOps engineers.
• Applied Gemini and OpenAI APIs to build GenAI-enabled assistants for telecom payment systems (PCI/DSS, ISO 20022/SWIFT) and 5G network slicing analysis.
• Leading and chairing DevOps and DevSecOps in multi-cloud at AT&T and Dish Networks had significant progress within the organization with 30% Operational costs.
• Leading and chairing ERP and CRM in OSS/BSS domain at T-Mobile led to a holistic design of best order orchestration, revenue reporting, service enablement, inventory management, billing and service assurance with 20% operational cost reduction.

In all of the above use cases in lab and prod, I managed different tangible Cloud/AI/Data engineering solution with a large number of architecture design and hands-on experience with tools and technics in hyper-scaler clouds, on-prem solutions.

NLP, RAG&LLM (different from hyperscaler cloud/llm providers), MLflow, Chatbots, LLM, VectorSearch db, Embeding, SemanticSearch, Retrieval, and PredictiveAI techniques. Different AI and ML tools and techniques not limited to Amazon Connect, AWS Bedrock, AWS SageMaker, AWS Kendra, Azure AI Suite, Azure AI search DB, Databricks, TensorFlow and PyTorch, and tools to deploy on different GPUs.

AI tools and platforms have you worked with:

AWS Bedrock, AWS Connect, AWS Kendra, AWS SageMaker, Databricks, Azure AI suite, Google Colab, Azure OpenAI, Azure Search, Tensorflow, Pytotch Python 3.2, OpenSearch, Elastic ELK, Grafana and W&B for observability.

At Dish Networks I was chairing ARB (Architecture Review Board) to review all the Cloud Network Architecture and Cloud security Hardening including the data privacy at transit and rest, hashing/encryption algorithms and the way top apply on Kubernetes control plane and data plane, Network security including DDoS prevention, intrusion prevention, Firewalls, routes restriction, and also application/API security like images and artifacts secret management, JWT, mTLS, SSO, tokenization and RBAC (Role Base Access Control) and ABAC (Attribute Base Access Control).
The challenges for me were:

1. Fine-grained roles defined in IAM (or similar identity providers and access level hierarchy in a complex organization
2. RoleBinding and Cluster RoleBinding in cloud environment
3. Network Security in a complex and multi-island organization (different network routes and firewall rules in different segments of the underlying networks managed by different people)
4. Complex anti-malware vulnerability testing and UAT in prod environments
5. Complexity in external secret managers in multi-cloud environments

At Ericsson and Dish Network I designed an OSS/BSS platform with comprehensive ERP and CRM solutions in entire T-Mobile Uprising and EchoStar large enterprises including Service Activation and enablement, Self-provisioning, Order Management, Order Orchestration, Charging and billing mediation, revenue assurance and reporting, Business Intelligence, Service Assurance, Organization Management, Inventory Management.

I made a risk assessment on AT&T prod environment to run an SDN controller based auto-scaling solution on an API gateway. I had least information about underlying infrastructure and firewall rules. I needed to extract some rules and put some pre-assumptions considering the risk of wrong assumptions, and how to mitigate this wrong assumptions. Full risk assessment in a CSV document with details of the each assignment is a required step before applying on the prod environment. I have implemented this methodology multiple times at Cisco Systems Canada, AT&T, Verizon, T-Mobile, Nokia and Dish Networks. Initially I have learnt this fast-paced/ambiguous environment with uncertainty.

Dish started my initiative in IaC for Telco Cloud (multi-cloud and not only AWS) for Autimation of Kubernetes based infrastructure bootstrapping (k8s control plane like GKE or EKS). I used 3gpp, Nephio, ONAL, CNCF, k8s.io and TMF in my initiative.

At Dish Network I was leading a team of 10 engineers as DevOps for CAF CICD (Cloud-Agnostic Framework) to fulfill the requirement I finalized from different application design team in my Wireless Cloud team. AWS and OCI were the pilot phase cloud platform used for this cloud agnostic framework.

I developed opensource tools and scaled on my work on all 3 layers of application, network, and infra. Using tools in category of LCM/CICD and Orchestration, up to management and FCAPS (observability tools) and also the testing tools, I always incorporated opensource standards and available tools/techniques to use the tool to take it to production.
That includes but not limited to Open5GC, OpenRAN, OpenStack, Openshift, NGINX, OpenDaylight, Kibana, Apache Kafka, Apache Heap, Tosca template, CSAR, HEAT template. This improved the operational efficiency by 35-40 percent and reduced Opex by 45 percent.